diff --git a/.github/workflows/mcb7-unified-runner-checks.yml b/.github/workflows/mcb7-unified-runner-checks.yml index ffc4c025..7bcffe4b 100644 --- a/.github/workflows/mcb7-unified-runner-checks.yml +++ b/.github/workflows/mcb7-unified-runner-checks.yml @@ -65,6 +65,7 @@ jobs: OPENAI_API_KEY: ci-placeholder MINICPM_API_KEY: ci-placeholder TYPESAFE_API_KEY: ci-placeholder + GEMINI_API_KEY: ci-placeholder run: | uv run model-bench models uv run model-bench tasks diff --git a/experiments/model-capability-benchmark/HOW_TO_USE.md b/experiments/model-capability-benchmark/HOW_TO_USE.md index eb7cc72a..c3b28b93 100644 --- a/experiments/model-capability-benchmark/HOW_TO_USE.md +++ b/experiments/model-capability-benchmark/HOW_TO_USE.md @@ -501,6 +501,99 @@ benchmark semantics. Start a new campaign/run instead. ## 12. Run all five core verticals for one model +### Consolidate runs by latest sample attempt + +`model-bench consolidate` writes a new immutable run from existing evidence, +without inference. Repeat `--run-dir` for every base/recovery run to include: + +```bash +uv run model-bench consolidate \ + --run-dir results/runs/base-run \ + --run-dir results/runs/recovery-run \ + --run-id model-consolidated-v1 \ + --run-group model-consolidated \ + --plan-only +``` + +Remove `--plan-only` to materialize the snapshot and report. The output must not +exist: use a new run ID for each new consolidation. Add all five base capability +runs to retain a complete model result, then include recovery runs as additional +sources. The source directory order does not affect selection. + +The logical key includes model, capability, dataset revision/split, sample ID, +task version, prompt version and suite version. Selection uses the latest state +timestamp; ties use source run ID, attempt and case ID in that order. A newer +invalid result replaces an older valid result. Inference and evaluation are copied +from the same attempt. No latest incomplete/failed attempt without paired evidence +is silently dropped or replaced by an older success: consolidate rejects it and +requires recovery first. Profiles, seeds, artifacts and dataset/task definitions +must be compatible. Sources are checksumed before and after reading. + +Each selected case retains its original run/case/attempt, generation settings, +timestamp and benchmark/runtime signatures. Aggregate metrics are recalculated on +the deduplicated cases; original runs are untouched. Mixed generation budgets are +explicit in the manifest and dashboard. Consolidated benchmark signatures are +separate from uniform-budget comparisons. Hardware/resource evidence is not +attributed to a single new execution of the consolidated cases. + +```bash +uv run model-bench project --rebuild --export-dashboard +uv run model-bench dashboard-build +``` + +Dashboard exports default to `--result-view latest`: the newest completed +consolidated result replaces the base cells for that model/capability/profile; +other models keep their ordinary CURRENT cells. Recovery subsets alone never +replace complete results. Consolidated runs are labeled in the dashboard. +To inspect only ordinary uniform-configuration results instead: + +```bash +uv run model-bench dashboard-data --result-view standard +uv run model-bench dashboard-build +``` + +The canonical SQL views `v_current_quality_results` and +`v_current_performance_results` remain restricted to standard runs. The dashboard's +latest selection is separately exposed as `v_latest_quality_results`. + +### Recover only invalid or failed cases with a different output budget + +Use `--rerun-errors-from` to select unsuccessful latest cases from a previous run. +This includes `completed` cases whose evaluation has `valid=false`, as well as +pipeline `failed` cases. Unlike `--retry-failures`, this creates a separate recovery +experiment and can change the token budget. A newer valid result for the same +model/capability/dataset/sample supersedes an earlier invalid result in the source +evidence. Repeat the source option to combine multiple run directories. + +```bash +export BENCHMARK_TIMEOUT_SECONDS=180 + +uv run model-bench run \ + --models "$MODEL" \ + --capabilities structured-output \ + --profile core \ + --run-group "${CAMPAIGN}-recovery-4096" \ + --run-id "${CAMPAIGN}-${MODEL}-structured-output-recovery-4096" \ + --rerun-errors-from "results/runs/${CAMPAIGN}-${MODEL}-structured-output-core" \ + --max-output-tokens 4096 \ + --plan-only +``` + +Remove `--plan-only` to execute the selection. The preview does not contact a +provider or create output files. Keep the source profile and seed; requested samples +must still exist in the current dataset selection. Use a new run ID and output +directory. Recovery manifests record source run IDs, state checksums, selected +logical cases and the generation override; the original evidence is untouched. +Repeating the same recovery command resumes it. If selection or token override +changes, use another new run ID. + +Recovery runs remain available in history but are excluded from canonical CURRENT +selection. Their scores measure recovery on an error-selected subset, not quality +on the complete benchmark. The existing `BENCHMARK_TIMEOUT_SECONDS` environment +variable controls provider request timeouts; more output tokens alone do not raise +the timeout. `--max-output-tokens` also works for ordinary full runs without editing +the shared suite configuration. + After you are comfortable running the first vertical manually: ~~~bash @@ -551,6 +644,55 @@ uv run model-bench dashboard-build ## 14. Move to the next model +### One public JSON file for every dashboard view + +Every dashboard export automatically replaces +`results/published/dashboard-data.json`. This is a single, versioned bundle with +`overview`, `capabilities`, `models` and `runs` maps: all routes, including frontier +and sensitivity (inside overview), can use the same snapshot. `generated_at_utc` +identifies when it was exported. The source is the selected analytics view, so +consolidated results and their mixed-budget labels are retained. + +Both commands refresh the public bundle automatically: + +```bash +uv run model-bench project --export-dashboard +uv run model-bench dashboard-data +``` + +The bundle is serialized completely and then atomically replaces the previous file. +Failed exports preserve the previous public file. Only files referenced by the +current export index are included; stale files are excluded. Known credential fields +and local filesystem paths are redacted, including serialized event metadata. +The original internal per-view exports remain available for local analysis. + +Host this file at a stable URL, for example `/data/dashboard-data.json`. To make the +existing dashboard load all views dynamically, set this **before** its module script: + +```html + +``` + +The application fetches once on startup with `cache: 'no-store'`, validates the +bundle envelope and initializes every route from it. A load error is displayed +explicitly, without falling back to fixture data. Existing embedded/offline builds +continue to work when no URL is configured. An external page can also consume it: + +```javascript +const response = await fetch('/data/dashboard-data.json', { cache: 'no-store' }); +if (!response.ok) throw new Error(`Dashboard data: HTTP ${response.status}`); +const snapshot = await response.json(); +const leaderboard = snapshot.overview.decision.model_summaries; +const capabilities = snapshot.capabilities; +const models = snapshot.models; +const runs = snapshot.runs; +``` + +Updating the local bundle does not upload it: the hosting/deployment pipeline must +sync `results/published/dashboard-data.json` to that URL. Serve JSON with compression +and, if hosted on a different origin, appropriate CORS headers. Reloading the page +loads the current snapshot; already-open pages do not poll automatically. + Example: ~~~bash diff --git a/experiments/model-capability-benchmark/README.md b/experiments/model-capability-benchmark/README.md index c89ed934..e90a2e43 100644 --- a/experiments/model-capability-benchmark/README.md +++ b/experiments/model-capability-benchmark/README.md @@ -165,9 +165,21 @@ New runs are stored by default under `results/runs/`, while `run_group` Korgis model residency is managed through its admin API; the Korgis server process itself is intentionally external to this benchmark. +`model-bench consolidate` creates an immutable latest-attempt result from base and +recovery runs, deduplicating logical samples and recalculating metrics without +inference. Dashboard exports prefer explicitly consolidated results by default and +label mixed budgets; `--result-view standard` retains the uniform-configuration +CURRENT view. Source evidence is preserved. See [the consolidation runbook](HOW_TO_USE.md#consolidate-runs-by-latest-sample-attempt). + See `MCB_7_UNIFIED_RUNNER.md` for architecture, failure semantics and the controlled real-provider validation runbook. +Dashboard exports also atomically update `results/published/dashboard-data.json`, +a public bundle containing every view's data. Set `window.__MCB_DATA_URL__` before +loading the dashboard to fetch that single file from a stable hosting URL. +See [the public-data runbook](HOW_TO_USE.md#one-public-json-file-for-every-dashboard-view) +for the schema, automatic refresh commands and hosting setup. + ## Capability report Every run now produces a neutral capability × model report without an overall score or diff --git a/experiments/model-capability-benchmark/dashboard/package.json b/experiments/model-capability-benchmark/dashboard/package.json index ab38a56b..5c7baa3f 100644 --- a/experiments/model-capability-benchmark/dashboard/package.json +++ b/experiments/model-capability-benchmark/dashboard/package.json @@ -10,7 +10,8 @@ "preview": "vite preview", "test:compare": "node --experimental-strip-types --test src/components/compareThreeLogic.test.ts src/components/hardwareEvidence.test.ts src/components/compareCockpitLogic.test.ts", "test:frontier": "node --experimental-strip-types --test src/components/frontierSelection.test.ts", - "test:guide": "node --experimental-strip-types --test src/datasetGuideData.test.ts" + "test:guide": "node --experimental-strip-types --test src/datasetGuideData.test.ts", + "test:bundle": "node --experimental-strip-types --test src/dashboardBundle.test.ts" }, "dependencies": { "lucide-react": "^1.47.0", diff --git a/experiments/model-capability-benchmark/dashboard/src/components/Shell.tsx b/experiments/model-capability-benchmark/dashboard/src/components/Shell.tsx index 8cbd8c8d..b5804b40 100644 --- a/experiments/model-capability-benchmark/dashboard/src/components/Shell.tsx +++ b/experiments/model-capability-benchmark/dashboard/src/components/Shell.tsx @@ -13,6 +13,7 @@ import { TrendingUp, } from 'lucide-react'; import type { ReactNode } from 'react'; +import { overview } from '../data'; import { navigate, usePathname } from '../router'; function Link({ @@ -123,7 +124,20 @@ export function Shell({ children }: { children: ReactNode }) { -
{children}
+
+ {overview.consolidations?.length ? ( +
+ Consolidated results · latest attempt per sample · + {overview.consolidations.some((row) => row.generation_policy.mixed_generation) + ? ' mixed generation budgets; not a uniform-budget comparison. ' + : ' combined source runs. '} + {[...new Set(overview.consolidations.map((row) => row.run_id))].map((runId) => ( + {runId} + ))} +
+ ) : null} + {children} +
); } diff --git a/experiments/model-capability-benchmark/dashboard/src/dashboardBundle.test.ts b/experiments/model-capability-benchmark/dashboard/src/dashboardBundle.test.ts new file mode 100644 index 00000000..7f02c1d2 --- /dev/null +++ b/experiments/model-capability-benchmark/dashboard/src/dashboardBundle.test.ts @@ -0,0 +1,21 @@ +import assert from 'node:assert/strict'; +import test from 'node:test'; +import { parseDashboardBundle } from './dashboardBundle.ts'; + +const valid = { + schema_version: '1', generated_at_utc: '2026-10-22T00:00:00Z', + overview: { models: [], capabilities: [], cells: [], runs: [] }, + capabilities: { cap: {} }, models: { model: {} }, runs: { run: {} }, +}; + +test('accepts one snapshot containing all route maps', () => { + assert.equal(parseDashboardBundle(valid), valid); +}); + +test('rejects unsupported, missing or malformed route data', () => { + for (const invalid of [null, {}, { ...valid, schema_version: '2' }, + { ...valid, models: [] }, { ...valid, runs: null }, { ...valid, capabilities: { cap: 'bad' } }, + { ...valid, overview: { ...valid.overview, cells: 'bad' } }]) { + assert.throws(() => parseDashboardBundle(invalid), /Unsupported or incomplete/); + } +}); diff --git a/experiments/model-capability-benchmark/dashboard/src/dashboardBundle.ts b/experiments/model-capability-benchmark/dashboard/src/dashboardBundle.ts new file mode 100644 index 00000000..bc31e1a5 --- /dev/null +++ b/experiments/model-capability-benchmark/dashboard/src/dashboardBundle.ts @@ -0,0 +1,19 @@ +import type { DashboardBundle } from './types'; + +function record(value: unknown): value is Record { + return typeof value === 'object' && value !== null && !Array.isArray(value); +} + +export function parseDashboardBundle(value: unknown): DashboardBundle { + if (!record(value) || value.schema_version !== '1' || typeof value.generated_at_utc !== 'string' + || !record(value.overview)) { + throw new Error('Unsupported or incomplete dashboard data file'); + } + const overview = value.overview; + if (!['models', 'capabilities', 'cells', 'runs'].every((key) => Array.isArray(overview[key])) + || !['capabilities', 'models', 'runs'].every((key) => record(value[key]) + && Object.values(value[key]).every(record))) { + throw new Error('Unsupported or incomplete dashboard data file'); + } + return value as unknown as DashboardBundle; +} diff --git a/experiments/model-capability-benchmark/dashboard/src/data.ts b/experiments/model-capability-benchmark/dashboard/src/data.ts index 6ee76afc..d4b4fa5e 100644 --- a/experiments/model-capability-benchmark/dashboard/src/data.ts +++ b/experiments/model-capability-benchmark/dashboard/src/data.ts @@ -2,6 +2,7 @@ import overviewFixture from './data/overview.fixture.json'; import capabilityFixture from './data/structured-output.fixture.json'; import type { CapabilityPayload, + DashboardBundle, ModelPayload, OverviewPayload, RunPayload, @@ -9,6 +10,8 @@ import type { declare global { interface Window { + __MCB_DATA_URL__?: string; + __MCB_DASHBOARD_DATA__?: DashboardBundle; __MCB_OVERVIEW__?: OverviewPayload; __MCB_CAPABILITY__?: CapabilityPayload; __MCB_CAPABILITIES__?: Record; @@ -18,15 +21,17 @@ declare global { } export const overview = - (window.__MCB_OVERVIEW__ ?? overviewFixture) as OverviewPayload; + (window.__MCB_DASHBOARD_DATA__?.overview ?? window.__MCB_OVERVIEW__ ?? overviewFixture) as OverviewPayload; export const capability = - (window.__MCB_CAPABILITY__ ?? capabilityFixture) as CapabilityPayload; + (window.__MCB_DASHBOARD_DATA__?.capabilities['structured-output'] + ?? Object.values(window.__MCB_DASHBOARD_DATA__?.capabilities ?? {})[0] + ?? window.__MCB_CAPABILITY__ ?? capabilityFixture) as CapabilityPayload; export const capabilityPayloads: Record = - window.__MCB_CAPABILITIES__ ?? { + window.__MCB_DASHBOARD_DATA__?.capabilities ?? window.__MCB_CAPABILITIES__ ?? { [capability.capability_id]: capability, }; -export const modelPayloads = window.__MCB_MODELS__ ?? {}; -export const runPayloads = window.__MCB_RUNS__ ?? {}; +export const modelPayloads = window.__MCB_DASHBOARD_DATA__?.models ?? window.__MCB_MODELS__ ?? {}; +export const runPayloads = window.__MCB_DASHBOARD_DATA__?.runs ?? window.__MCB_RUNS__ ?? {}; diff --git a/experiments/model-capability-benchmark/dashboard/src/main.tsx b/experiments/model-capability-benchmark/dashboard/src/main.tsx index ed3b67f5..34a0261e 100644 --- a/experiments/model-capability-benchmark/dashboard/src/main.tsx +++ b/experiments/model-capability-benchmark/dashboard/src/main.tsx @@ -1,10 +1,23 @@ import { StrictMode } from 'react'; import { createRoot } from 'react-dom/client'; -import { App } from './App'; +import { parseDashboardBundle } from './dashboardBundle'; import './styles.css'; -createRoot(document.getElementById('root')!).render( - - - , -); +async function start() { + if (window.__MCB_DATA_URL__) { + const response = await fetch(window.__MCB_DATA_URL__, { cache: 'no-store' }); + if (!response.ok) throw new Error(`Dashboard data: HTTP ${response.status}`); + window.__MCB_DASHBOARD_DATA__ = parseDashboardBundle(await response.json()); + } + // Import after loading data: every route uses the same complete snapshot. + const { App } = await import('./App'); + createRoot(document.getElementById('root')!).render( + , + ); +} + +start().catch((error: unknown) => { + const root = document.getElementById('root')!; + root.setAttribute('role', 'alert'); + root.textContent = error instanceof Error ? error.message : 'Unable to load dashboard data'; +}); diff --git a/experiments/model-capability-benchmark/dashboard/src/pages/ExecutivePages.tsx b/experiments/model-capability-benchmark/dashboard/src/pages/ExecutivePages.tsx index a3675545..b563a65b 100644 --- a/experiments/model-capability-benchmark/dashboard/src/pages/ExecutivePages.tsx +++ b/experiments/model-capability-benchmark/dashboard/src/pages/ExecutivePages.tsx @@ -1221,7 +1221,7 @@ function ReliabilityBars({ models }: { models: DecisionModelSummary[] }) { return (
- {model.model_key} + {model.model_key}{model.result_policy === 'consolidated-latest-attempt' ? ' · consolidated' : ''} {model.quality_coverage_complete ? 'Complete quality coverage' : 'Partial quality coverage'} diff --git a/experiments/model-capability-benchmark/dashboard/src/types.ts b/experiments/model-capability-benchmark/dashboard/src/types.ts index 5dd074ac..e4951b6f 100644 --- a/experiments/model-capability-benchmark/dashboard/src/types.ts +++ b/experiments/model-capability-benchmark/dashboard/src/types.ts @@ -90,6 +90,7 @@ export interface QualityPolicy { } export interface DecisionModelSummary extends DashboardModel { + result_policy?: 'standard' | 'consolidated-latest-attempt'; quality_policy_id: string; overall_quality_score: number | null; quality_coverage_capabilities: number; @@ -280,6 +281,13 @@ export interface SensitivityPayload { export interface OverviewPayload { schema_version: string; + result_view?: 'latest' | 'standard'; + consolidations?: Array<{ + run_id: string; + model_key: string; + capability_id: string; + generation_policy: { mixed_generation?: boolean; generation_configs?: unknown[] }; + }>; fixture?: boolean; models: DashboardModel[]; capabilities: string[]; @@ -290,6 +298,15 @@ export interface OverviewPayload { sensitivity?: SensitivityPayload; } +export interface DashboardBundle { + schema_version: '1'; + generated_at_utc: string; + overview: OverviewPayload; + capabilities: Record; + models: Record; + runs: Record; +} + export interface FamilyBreakdown { model_key: string; family: string; diff --git a/experiments/model-capability-benchmark/models.yaml b/experiments/model-capability-benchmark/models.yaml index 2e6add30..dbfd8f34 100644 --- a/experiments/model-capability-benchmark/models.yaml +++ b/experiments/model-capability-benchmark/models.yaml @@ -144,6 +144,19 @@ runtimes: models: + lfm2.5-24b-a2b-ud-q3-k-xl: + model_id: unsloth/LFM2.5-24B-A2B-GGUF + runtime_model_id: lfm2.5-24b-a2b-ud-q3-k-xl + runtime: korgis-local + family: lfm2.5 + parameters_b: 24 + artifact: + format: gguf + quantization: UD-Q3_K_XL + metadata: + resolved_by: korgis-model-registry + tags: [local, open-weight, moe] + qwen3.5-0.8b-q4km: model_id: unsloth/qwen3.5-0.8b runtime_model_id: qwen3.5-0.8b-q4km diff --git a/experiments/model-capability-benchmark/results/published/dashboard-data.json b/experiments/model-capability-benchmark/results/published/dashboard-data.json new file mode 100644 index 00000000..c8b949ac --- /dev/null +++ b/experiments/model-capability-benchmark/results/published/dashboard-data.json @@ -0,0 +1 @@ +{"capabilities":{"intent-classification":{"benchmark_signatures":["sha256:benchmark:718e761e4ba3410c289a5bba91a6474827caaa8da30e644b3df925ba4d16fbfc","sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","sha256:consolidated:42ab94943101512840504e431240a2dc1e9c2c51d6e055498c7b6b44c803f515"],"capability_id":"intent-classification","cases":[],"cells":[{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-06T10:03:19.101787+00:00","deployment":"local","effective_model_id":"bonsai-27b-q10","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.5974025974025974,"profile":"core","provider_key":"korgis","quantization":"Q1_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"bonsai-27b-q10","process_cpu_percent_avg":4.141997751956015,"process_rss_bytes_avg":5899303355.703257,"process_rss_bytes_peak":10491068416.0,"run_id":"2026-10-core-v1-bonsai-27b-q10-intent-classification-core","sample_count":4416,"sample_interval_ms":317.0853588508965,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4949098496.0},"run_id":"2026-10-core-v1-bonsai-27b-q10-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:718e761e4ba3410c289a5bba91a6474827caaa8da30e644b3df925ba4d16fbfc","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T15:04:59.981803+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"2e7b64ca38a2db66a9d8b36910d2b2ac5da0c7ce","model_id":"unsloth/qwen3.5-2b","model_key":"decisio-qwen3.5-2b-q4km","model_signature":"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.5974025974025974,"profile":"smoke","provider_key":"decisio","quantization":"Q4_K_M","resource_summary":null,"run_id":"16cd6c17-c9de-422d-8583-1a17959bdf6b","runtime_key":"decisio-local","sample_count":77,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T15:26:12.447493+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"2e7b64ca38a2db66a9d8b36910d2b2ac5da0c7ce","model_id":"unsloth/qwen3.5-2b","model_key":"decisio-qwen3.5-2b-q4km","model_signature":"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6038961038961039,"profile":"core","provider_key":"decisio","quantization":"Q4_K_M","resource_summary":null,"run_id":"1e48c427-4e75-4ace-a465-2aec7e2751cd","runtime_key":"decisio-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:718e761e4ba3410c289a5bba91a6474827caaa8da30e644b3df925ba4d16fbfc","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T15:11:29.046506+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"2e7b64ca38a2db66a9d8b36910d2b2ac5da0c7ce","model_id":"unsloth/qwen3.5-4b","model_key":"decisio-qwen3.5-4b-q4km","model_signature":"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6753246753246753,"profile":"smoke","provider_key":"decisio","quantization":"Q4_K_M","resource_summary":null,"run_id":"a97505cf-abc6-4782-bf99-38de2a4492b7","runtime_key":"decisio-local","sample_count":77,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T15:33:33.988827+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"2e7b64ca38a2db66a9d8b36910d2b2ac5da0c7ce","model_id":"unsloth/qwen3.5-4b","model_key":"decisio-qwen3.5-4b-q4km","model_signature":"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.7012987012987013,"profile":"core","provider_key":"decisio","quantization":"Q4_K_M","resource_summary":null,"run_id":"b906e312-aa49-4397-b60e-a1c3ad6a8f87","runtime_key":"decisio-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:718e761e4ba3410c289a5bba91a6474827caaa8da30e644b3df925ba4d16fbfc","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T16:17:35.400024+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"0065acc5513f65c3e8bbdf6899cf55e286dfe938","model_id":"qwen/qwen3.5-9b","model_key":"decisio-qwen3.5-9b-q4km","model_signature":"sha256:model:ae0099eeb8925bc828c92072ba2f486986899854d2d3de76e5689cd383c783e3","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6623376623376623,"profile":"smoke","provider_key":"decisio","quantization":"Q4_K_M","resource_summary":null,"run_id":"e9c4ec37-433f-4b06-b397-9d9b3a4f7480","runtime_key":"decisio-local","sample_count":77,"task_id":"intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T13:59:58.574010+00:00","deployment":"api","effective_model_id":"gemini-3.5-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.5","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.0,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core","runtime_key":"gemini-api","sample_count":154,"task_id":"intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T13:39:20.878424+00:00","deployment":"api","effective_model_id":"gemini-3.8-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.8","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.8-flash","model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.0,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core","runtime_key":"gemini-api","sample_count":154,"task_id":"intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-06T12:22:05.300981+00:00","deployment":"api","effective_model_id":"gpt-5.6-luna","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"11b00f55c776ec89ddee7cd2dd1898797dde4055","model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.8246753246753247,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-intent-classification-core","runtime_key":"openai-api","sample_count":154,"task_id":"intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T13:18:34.815598+00:00","deployment":"api","effective_model_id":"gpt-6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","experiment_kind":"standard","failure_count":0,"family":"gpt-6","generation_policy":{},"git_commit":"7f796c10d99e934da6417ea54405fbf9352ad399","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.7987012987012987,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"gpt-6-luna-core-gpt-6-luna-intent-classification-core","runtime_key":"openai-api","sample_count":154,"task_id":"intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-06T12:50:59.652242+00:00","deployment":"api","effective_model_id":"jev-1.13.0","execution_signature":"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3","experiment_kind":"standard","failure_count":0,"family":"jev","generation_policy":{},"git_commit":"72697127df45f8a3abc98b0b7bdf6fd53a0b5c98","model_id":"jev-1.13.0","model_key":"jev-1.13.0","model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.7467532467532467,"profile":"core","provider_key":"typesafe","quantization":null,"resource_summary":null,"run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","runtime_key":"typesafe-api","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":10947100928,"benchmark_signature":"sha256:consolidated:42ab94943101512840504e431240a2dc1e9c2c51d6e055498c7b6b44c803f515","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-22T13:12:52.242438+00:00","deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","experiment_kind":"consolidated","failure_count":0,"family":"lfm2.5","generation_policy":{"generation_configs":[{"extra":{},"max_output_tokens":2048,"seed":42,"stop":[],"temperature":0.0},{"extra":{},"max_output_tokens":4096,"seed":42,"stop":[],"temperature":0.0}],"mixed_generation":true},"git_commit":null,"model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.7012987012987013,"profile":"core","provider_key":"korgis","quantization":"UD-Q3_K_XL","resource_summary":null,"run_id":"lfm25-consolidated-latest","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T12:03:20.624151+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.5584415584415584,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b","process_cpu_percent_avg":5.91751681481258,"process_rss_bytes_avg":5777497087.999999,"process_rss_bytes_peak":10717691904.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-intent-classification-core","sample_count":1252,"sample_interval_ms":293.3800480326384,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6120767488.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T13:24:18.004156+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b-q8","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.551948051948052,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b-q8","process_cpu_percent_avg":4.532544993975123,"process_rss_bytes_avg":5542900553.420978,"process_rss_bytes_peak":9062940672.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core","sample_count":1798,"sample_interval_ms":315.6007340511925,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5441241088.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T13:37:29.943647+00:00","deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.2792207792207792,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-0.8b-q4km","process_cpu_percent_avg":12.95914646037019,"process_rss_bytes_avg":4417102107.705627,"process_rss_bytes_peak":7946436608.0,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core","sample_count":462,"sample_interval_ms":199.48604018792173,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":10223632384.0},"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":3775709216,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T11:24:59.796058+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-bf16","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","parameters_b":2.0,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.577922077922078,"profile":"core","provider_key":"korgis","quantization":"BF16","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","model_key":"qwen3.5-2b-bf16","process_cpu_percent_avg":5.488711560059013,"process_rss_bytes_avg":5427256115.200001,"process_rss_bytes_peak":7206486016.0,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core","sample_count":618,"sample_interval_ms":285.9658953851292,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5988106240.0},"run_id":"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T11:25:37.855179+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.577922077922078,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-2b-q4km","process_cpu_percent_avg":9.421328984702535,"process_rss_bytes_avg":2155211762.7012987,"process_rss_bytes_peak":3131637760.0,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core","sample_count":616,"sample_interval_ms":260.0283379352414,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6455427072.0},"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T16:36:39.120802+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"cb061291446b8dbd79f524a4fefcd4a4406e2a0d","model_id":"unsloth/qwen3.5-4b","model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6688311688311688,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-4b-q4km","process_cpu_percent_avg":6.008969494809468,"process_rss_bytes_avg":5011243553.1201,"process_rss_bytes_peak":7911129088.0,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":1067,"sample_interval_ms":281.19401557671836,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5343313920.0},"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":4482403488,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T11:37:13.963536+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q8","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","parameters_b":4.0,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6493506493506493,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","model_key":"qwen3.5-4b-q8","process_cpu_percent_avg":5.113588871923991,"process_rss_bytes_avg":7039335235.285094,"process_rss_bytes_peak":10390700032.0,"run_id":"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core","sample_count":1041,"sample_interval_ms":311.43875954839615,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5581979648.0},"run_id":"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T12:46:44.225239+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6948051948051948,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-9b-q4km","process_cpu_percent_avg":3.9138417761512048,"process_rss_bytes_avg":6965862817.821014,"process_rss_bytes_peak":10559422464.0,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core","sample_count":1620,"sample_interval_ms":303.4210088575144,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5489934336.0},"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":4121781472,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T11:55:33.171703+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-ud-q2kxl","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"qwen/qwen3.5-9b-ud-q2kxl","model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","parameters_b":9.0,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6883116883116883,"profile":"core","provider_key":"korgis","quantization":"UD-Q2_K_XL","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","model_key":"qwen3.5-9b-ud-q2kxl","process_cpu_percent_avg":3.288803542984178,"process_rss_bytes_avg":6954068738.250033,"process_rss_bytes_peak":11224416256.0,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","sample_count":1539,"sample_interval_ms":314.5480896344338,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5547704320.0},"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":6192222208,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-06T14:37:14.912068+00:00","deployment":"local","effective_model_id":"qwen3.8-27b-iq1s","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","experiment_kind":"standard","failure_count":0,"family":"qwen3.8","generation_policy":{},"git_commit":"640138c3429882a42797d28074f98d3fea98a63b","model_id":"Qwen/Qwen3.8-27B","model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","parameters_b":27.0,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.5844155844155844,"profile":"core","provider_key":"korgis","quantization":"UD-IQ1_S","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","model_key":"qwen3.8-27b-iq1s","process_cpu_percent_avg":3.5068517339810183,"process_rss_bytes_avg":8038410776.168479,"process_rss_bytes_peak":12158550016.0,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","sample_count":4560,"sample_interval_ms":309.607195745226,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5166546944.0},"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-06T09:08:01.471590+00:00","deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"xhtoken/spark-x2.5-4b","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6493506493506493,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"spark-x2.5-4b-q4km","process_cpu_percent_avg":5.594450252441077,"process_rss_bytes_avg":3797578739.069748,"process_rss_bytes_peak":6051266560.0,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core","sample_count":1308,"sample_interval_ms":310.7691591083291,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5891014656.0},"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"}],"decision":{"capability_summaries":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":8698.848109240469,"latency_p50_ms":8900.03939550661,"latency_p95_ms":9739.512333347375,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","normalized_quality_score":59.74025974025974,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1529,"primary_metric":"accuracy","primary_value":0.5974025974025974,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-bonsai-27b-q10-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","failure_count":1,"failure_rate":0.006493506493506494,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1155.2659180455662,"latency_p50_ms":1149.7726044981391,"latency_p95_ms":1208.5230768483598,"model_key":"decisio-qwen3.5-2b-q4km","model_signature":"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95","normalized_quality_score":60.3896103896104,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1203,"primary_metric":"accuracy","primary_value":0.6038961038961039,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"1e48c427-4e75-4ace-a465-2aec7e2751cd","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","failure_count":1,"failure_rate":0.006493506493506494,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2834.3784266303433,"latency_p50_ms":2828.9587710023625,"latency_p95_ms":2931.071354152664,"model_key":"decisio-qwen3.5-4b-q4km","model_signature":"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a","normalized_quality_score":70.12987012987013,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1192,"primary_metric":"accuracy","primary_value":0.7012987012987013,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"b906e312-aa49-4397-b60e-a1c3ad6a8f87","sample_count":154},{"benchmark_signature":"sha256:benchmark:718e761e4ba3410c289a5bba91a6474827caaa8da30e644b3df925ba4d16fbfc","capability_id":"intent-classification","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","failure_count":0,"failure_rate":0.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":5000.302291662143,"latency_p50_ms":4986.015040994971,"latency_p95_ms":5194.435066799633,"model_key":"decisio-qwen3.5-9b-q4km","model_signature":"sha256:model:ae0099eeb8925bc828c92072ba2f486986899854d2d3de76e5689cd383c783e3","normalized_quality_score":66.23376623376623,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":603,"primary_metric":"accuracy","primary_value":0.6623376623376623,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"run_id":"e9c4ec37-433f-4b06-b397-9d9b3a4f7480","sample_count":77},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":154,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":131.37987608282052,"latency_p50_ms":127.4633540015202,"latency_p95_ms":156.07859991432633,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","normalized_quality_score":0.0,"observed_case_count":154,"observed_quality_latency_pareto":true,"output_tokens_total":null,"primary_metric":"accuracy","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":154,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":130.65657031785867,"latency_p50_ms":127.60222947690636,"latency_p95_ms":170.61041467823082,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","normalized_quality_score":0.0,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":null,"primary_metric":"accuracy","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.8-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","failure_count":4,"failure_rate":0.025974025974025976,"input_tokens_total":147621,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1730.830770566101,"latency_p50_ms":1359.8202709981706,"latency_p95_ms":4205.3196875946005,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","normalized_quality_score":82.46753246753246,"observed_case_count":154,"observed_quality_latency_pareto":true,"output_tokens_total":4896,"primary_metric":"accuracy","primary_value":0.8246753246753247,"provider_cost_coverage_rate":0.974025974025974,"provider_cost_known":true,"provider_cost_observed_total_usd":0.035399400000000004,"provider_cost_per_1k_cases_usd":0.23599600000000004,"provider_cost_per_case_usd":0.00023599600000000004,"provider_cost_priced_cases":150,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":0,"failure_rate":0.0,"input_tokens_total":151555,"known_provider_cost_quality_pareto":true,"latency_mean_ms":2025.4564632668444,"latency_p50_ms":1870.263104501646,"latency_p95_ms":3181.1156249023043,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","normalized_quality_score":79.87012987012987,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":8558,"primary_metric":"accuracy","primary_value":0.7987012987012987,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0405806,"provider_cost_per_1k_cases_usd":0.26351038961038964,"provider_cost_per_case_usd":0.00026351038961038963,"provider_cost_priced_cases":154,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":154,"provider_cost_total_usd":0.0405806,"run_id":"gpt-6-luna-core-gpt-6-luna-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3","failure_count":0,"failure_rate":0.0,"input_tokens_total":525524,"known_provider_cost_quality_pareto":true,"latency_mean_ms":254.11093069761864,"latency_p50_ms":243.48435449792305,"latency_p95_ms":309.60259374624,"model_key":"jev-1.13.0","model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","normalized_quality_score":74.67532467532467,"observed_case_count":308,"observed_quality_latency_pareto":true,"output_tokens_total":255096,"primary_metric":"accuracy","primary_value":0.7467532467532467,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.022072008,"provider_cost_per_1k_cases_usd":0.07166236363636364,"provider_cost_per_case_usd":7.166236363636363e-05,"provider_cost_priced_cases":308,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"jev","processing":"standard","source":"TypeSafe AI public pricing: $42 per billion input tokens; output free.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":308,"provider_cost_total_usd":0.022072008,"run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:consolidated:42ab94943101512840504e431240a2dc1e9c2c51d6e055498c7b6b44c803f515","capability_id":"intent-classification","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","failure_count":6,"failure_rate":0.03896103896103896,"input_tokens_total":96208,"known_provider_cost_quality_pareto":false,"latency_mean_ms":23087.311111795778,"latency_p50_ms":16176.900271035265,"latency_p95_ms":70185.2069376211,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","normalized_quality_score":70.12987012987013,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":159929,"primary_metric":"accuracy","primary_value":0.7012987012987013,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"lfm25-consolidated-latest","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":97089,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2025.159793870857,"latency_p50_ms":2010.3751250062487,"latency_p95_ms":2163.9121060507023,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","normalized_quality_score":55.84415584415584,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1669,"primary_metric":"accuracy","primary_value":0.5584415584415584,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":97089,"known_provider_cost_quality_pareto":false,"latency_mean_ms":3292.909132578643,"latency_p50_ms":3393.2761040050536,"latency_p95_ms":4267.191872800322,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","normalized_quality_score":55.1948051948052,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1678,"primary_metric":"accuracy","primary_value":0.551948051948052,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":342.47120458998586,"latency_p50_ms":339.46391649806174,"latency_p95_ms":371.62911084524234,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","normalized_quality_score":27.92207792207792,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1492,"primary_metric":"accuracy","primary_value":0.2792207792207792,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":785.6293642306065,"latency_p50_ms":779.2861880152486,"latency_p95_ms":870.3237018693471,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","normalized_quality_score":57.7922077922078,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1509,"primary_metric":"accuracy","primary_value":0.577922077922078,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":699.1592043047497,"latency_p50_ms":688.2556045020465,"latency_p95_ms":780.7678268472954,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","normalized_quality_score":57.7922077922078,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1500,"primary_metric":"accuracy","primary_value":0.577922077922078,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1596.720603324247,"latency_p50_ms":1580.0608540012036,"latency_p95_ms":1735.5302432952158,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","normalized_quality_score":66.88311688311688,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1508,"primary_metric":"accuracy","primary_value":0.6688311688311688,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1682.3662307401423,"latency_p50_ms":1674.4573339819908,"latency_p95_ms":1877.7022937458241,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","normalized_quality_score":64.93506493506493,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1515,"primary_metric":"accuracy","primary_value":0.6493506493506493,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2811.8513665976934,"latency_p50_ms":2785.3132915042806,"latency_p95_ms":3057.8174914451665,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","normalized_quality_score":69.48051948051948,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1507,"primary_metric":"accuracy","primary_value":0.6948051948051948,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2741.8135949013626,"latency_p50_ms":2726.615708990721,"latency_p95_ms":2938.3117294928525,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","normalized_quality_score":68.83116883116884,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1499,"primary_metric":"accuracy","primary_value":0.6883116883116883,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","failure_count":2,"failure_rate":0.012987012987012988,"input_tokens_total":84820,"known_provider_cost_quality_pareto":false,"latency_mean_ms":8791.119202123857,"latency_p50_ms":8568.107791492366,"latency_p95_ms":9098.29579560901,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","normalized_quality_score":58.44155844155844,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1449,"primary_metric":"accuracy","primary_value":0.5844155844155844,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":99556,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2224.7655532798503,"latency_p50_ms":2207.8276665051817,"latency_p95_ms":2593.4791601015604,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","normalized_quality_score":64.93506493506493,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":2139,"primary_metric":"accuracy","primary_value":0.6493506493506493,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core","sample_count":154}],"dataset_summaries":[{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":8698.848109240469,"latency_p50_ms":8900.03939550661,"latency_p95_ms":9739.512333347375,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","normalized_quality_score":59.74025974025974,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1529,"primary_metric":"accuracy","primary_value":0.5974025974025974,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":1,"failure_rate":0.006493506493506494,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1155.2659180455662,"latency_p50_ms":1149.7726044981391,"latency_p95_ms":1208.5230768483598,"model_key":"decisio-qwen3.5-2b-q4km","model_signature":"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95","normalized_quality_score":60.3896103896104,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1203,"primary_metric":"accuracy","primary_value":0.6038961038961039,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":1,"failure_rate":0.006493506493506494,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2834.3784266303433,"latency_p50_ms":2828.9587710023625,"latency_p95_ms":2931.071354152664,"model_key":"decisio-qwen3.5-4b-q4km","model_signature":"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a","normalized_quality_score":70.12987012987013,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1192,"primary_metric":"accuracy","primary_value":0.7012987012987013,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":5000.302291662143,"latency_p50_ms":4986.015040994971,"latency_p95_ms":5194.435066799633,"model_key":"decisio-qwen3.5-9b-q4km","model_signature":"sha256:model:ae0099eeb8925bc828c92072ba2f486986899854d2d3de76e5689cd383c783e3","normalized_quality_score":66.23376623376623,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":603,"primary_metric":"accuracy","primary_value":0.6623376623376623,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":154,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":131.37987608282052,"latency_p50_ms":127.4633540015202,"latency_p95_ms":156.07859991432633,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","normalized_quality_score":0.0,"observed_case_count":154,"observed_quality_latency_pareto":true,"output_tokens_total":null,"primary_metric":"accuracy","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":154,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":130.65657031785867,"latency_p50_ms":127.60222947690636,"latency_p95_ms":170.61041467823082,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","normalized_quality_score":0.0,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":null,"primary_metric":"accuracy","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.8-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":4,"failure_rate":0.025974025974025976,"input_tokens_total":147621,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1730.830770566101,"latency_p50_ms":1359.8202709981706,"latency_p95_ms":4205.3196875946005,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","normalized_quality_score":82.46753246753246,"observed_case_count":154,"observed_quality_latency_pareto":true,"output_tokens_total":4896,"primary_metric":"accuracy","primary_value":0.8246753246753247,"provider_cost_coverage_rate":0.974025974025974,"provider_cost_known":true,"provider_cost_observed_total_usd":0.035399400000000004,"provider_cost_per_1k_cases_usd":0.23599600000000004,"provider_cost_per_case_usd":0.00023599600000000004,"provider_cost_priced_cases":150,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":151555,"known_provider_cost_quality_pareto":true,"latency_mean_ms":2025.4564632668444,"latency_p50_ms":1870.263104501646,"latency_p95_ms":3181.1156249023043,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","normalized_quality_score":79.87012987012987,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":8558,"primary_metric":"accuracy","primary_value":0.7987012987012987,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0405806,"provider_cost_per_1k_cases_usd":0.26351038961038964,"provider_cost_per_case_usd":0.00026351038961038963,"provider_cost_priced_cases":154,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":154,"provider_cost_total_usd":0.0405806,"sample_count":154},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":525524,"known_provider_cost_quality_pareto":true,"latency_mean_ms":254.11093069761864,"latency_p50_ms":243.48435449792305,"latency_p95_ms":309.60259374624,"model_key":"jev-1.13.0","model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","normalized_quality_score":74.67532467532467,"observed_case_count":308,"observed_quality_latency_pareto":true,"output_tokens_total":255096,"primary_metric":"accuracy","primary_value":0.7467532467532467,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.022072008,"provider_cost_per_1k_cases_usd":0.07166236363636364,"provider_cost_per_case_usd":7.166236363636363e-05,"provider_cost_priced_cases":308,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"jev","processing":"standard","source":"TypeSafe AI public pricing: $42 per billion input tokens; output free.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":308,"provider_cost_total_usd":0.022072008,"sample_count":308},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":6,"failure_rate":0.03896103896103896,"input_tokens_total":96208,"known_provider_cost_quality_pareto":false,"latency_mean_ms":23087.311111795778,"latency_p50_ms":16176.900271035265,"latency_p95_ms":70185.2069376211,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","normalized_quality_score":70.12987012987013,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":159929,"primary_metric":"accuracy","primary_value":0.7012987012987013,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":97089,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2025.159793870857,"latency_p50_ms":2010.3751250062487,"latency_p95_ms":2163.9121060507023,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","normalized_quality_score":55.84415584415584,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1669,"primary_metric":"accuracy","primary_value":0.5584415584415584,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":97089,"known_provider_cost_quality_pareto":false,"latency_mean_ms":3292.909132578643,"latency_p50_ms":3393.2761040050536,"latency_p95_ms":4267.191872800322,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","normalized_quality_score":55.1948051948052,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1678,"primary_metric":"accuracy","primary_value":0.551948051948052,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":342.47120458998586,"latency_p50_ms":339.46391649806174,"latency_p95_ms":371.62911084524234,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","normalized_quality_score":27.92207792207792,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1492,"primary_metric":"accuracy","primary_value":0.2792207792207792,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":785.6293642306065,"latency_p50_ms":779.2861880152486,"latency_p95_ms":870.3237018693471,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","normalized_quality_score":57.7922077922078,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1509,"primary_metric":"accuracy","primary_value":0.577922077922078,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":699.1592043047497,"latency_p50_ms":688.2556045020465,"latency_p95_ms":780.7678268472954,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","normalized_quality_score":57.7922077922078,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1500,"primary_metric":"accuracy","primary_value":0.577922077922078,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1596.720603324247,"latency_p50_ms":1580.0608540012036,"latency_p95_ms":1735.5302432952158,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","normalized_quality_score":66.88311688311688,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1508,"primary_metric":"accuracy","primary_value":0.6688311688311688,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1682.3662307401423,"latency_p50_ms":1674.4573339819908,"latency_p95_ms":1877.7022937458241,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","normalized_quality_score":64.93506493506493,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1515,"primary_metric":"accuracy","primary_value":0.6493506493506493,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2811.8513665976934,"latency_p50_ms":2785.3132915042806,"latency_p95_ms":3057.8174914451665,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","normalized_quality_score":69.48051948051948,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1507,"primary_metric":"accuracy","primary_value":0.6948051948051948,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2741.8135949013626,"latency_p50_ms":2726.615708990721,"latency_p95_ms":2938.3117294928525,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","normalized_quality_score":68.83116883116884,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1499,"primary_metric":"accuracy","primary_value":0.6883116883116883,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":2,"failure_rate":0.012987012987012988,"input_tokens_total":84820,"known_provider_cost_quality_pareto":false,"latency_mean_ms":8791.119202123857,"latency_p50_ms":8568.107791492366,"latency_p95_ms":9098.29579560901,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","normalized_quality_score":58.44155844155844,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1449,"primary_metric":"accuracy","primary_value":0.5844155844155844,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":99556,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2224.7655532798503,"latency_p50_ms":2207.8276665051817,"latency_p95_ms":2593.4791601015604,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","normalized_quality_score":64.93506493506493,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":2139,"primary_metric":"accuracy","primary_value":0.6493506493506493,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154}]},"disagreements":[],"family_breakdown":[],"schema_version":"2"},"mathematical-reasoning":{"benchmark_signatures":["sha256:benchmark:382be54f0e8c250212fa1d53656a34eb22335c491d7aa5b1ba59e927a0c58dea","sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","sha256:benchmark:dca92e0e83499d6053d9585c0bdee4321b7ec9651563312492a659578feb94ee","sha256:consolidated:20f65d6394555078bc9dce0c1393e10d963aecb84ce9344ece12046a78d42de3"],"capability_id":"mathematical-reasoning","cases":[],"cells":[{"artifact_format":"gguf","artifact_size_bytes":3803452480,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"bonsai-27b-q10","execution_signature":"sha256:execution:80688f79e2c27746cabb9c8e268a98d63ee3d732cb5c73d70fc581a1958017fc","experiment_kind":"standard","failure_count":0,"family":"bonsai","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","parameters_b":27.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.475,"profile":"core","provider_key":"korgis","quantization":"Q1_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:80688f79e2c27746cabb9c8e268a98d63ee3d732cb5c73d70fc581a1958017fc","model_key":"bonsai-27b-q10","process_cpu_percent_avg":10.004219049002119,"process_rss_bytes_avg":7178716753.609695,"process_rss_bytes_peak":9109520384.0,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":342,"sample_interval_ms":299.5298028563516,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5797576704.0},"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"api","effective_model_id":"gemini-3.5-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.5","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.125,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"gemini-api","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T15:34:31.093143+00:00","deployment":"api","effective_model_id":"gpt-5.6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","experiment_kind":"standard","failure_count":0,"family":"gpt-5.6","generation_policy":{},"git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.675,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","runtime_key":"openai-api","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T15:34:31.093143+00:00","deployment":"api","effective_model_id":"gpt-6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","experiment_kind":"standard","failure_count":0,"family":"gpt-6","generation_policy":{},"git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.7,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","runtime_key":"openai-api","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":10947100928,"benchmark_signature":"sha256:consolidated:20f65d6394555078bc9dce0c1393e10d963aecb84ce9344ece12046a78d42de3","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-22T13:12:52.242438+00:00","deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","experiment_kind":"consolidated","failure_count":0,"family":"lfm2.5","generation_policy":{"generation_configs":[{"extra":{},"max_output_tokens":2048,"seed":42,"stop":[],"temperature":0.0},{"extra":{},"max_output_tokens":4096,"seed":42,"stop":[],"temperature":0.0}],"mixed_generation":true},"git_commit":null,"model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.95,"profile":"core","provider_key":"korgis","quantization":"UD-Q3_K_XL","resource_summary":null,"run_id":"lfm25-consolidated-latest","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:27:15.751735+00:00","deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","experiment_kind":"standard","failure_count":0,"family":"limite","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","parameters_b":1.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.9,"profile":"core","provider_key":"transformers","quantization":null,"resource_summary":null,"run_id":"16a842b0-5dcc-47b5-8808-850b2ad1ac10","runtime_key":"transformers-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:382be54f0e8c250212fa1d53656a34eb22335c491d7aa5b1ba59e927a0c58dea","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T14:59:31.170321+00:00","deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","experiment_kind":"standard","failure_count":0,"family":"limite","generation_policy":{},"git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65","parameters_b":1.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.7222222222222222,"profile":"extended","provider_key":"transformers","quantization":null,"resource_summary":null,"run_id":"2533bf73-b012-4d12-a6b6-9b6e4334dae2","runtime_key":"transformers-local","sample_count":72,"task_id":"mathematical-reasoning"},{"artifact_format":"safetensors","artifact_size_bytes":416,"benchmark_signature":"sha256:benchmark:dca92e0e83499d6053d9585c0bdee4321b7ec9651563312492a659578feb94ee","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-20T20:00:36.317482+00:00","deployment":"local","effective_model_id":"limite-1b-violetto-mlx","execution_signature":"sha256:execution:f4c681453ad106c8c878d601ee4a0ee6335674e81abaf3f7ba5297ce35841db0","experiment_kind":"standard","failure_count":0,"family":"limite","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"gabe-rbo/limite-1b-violetto-mlx","model_key":"limite-1b-violetto-mlx","model_signature":"sha256:model:fddf01c0845257f40bbb635a92054df8eb22dacd450b827a951962e906d6513b","parameters_b":1.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.0,"profile":"smoke","provider_key":"korgis","quantization":null,"resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:f4c681453ad106c8c878d601ee4a0ee6335674e81abaf3f7ba5297ce35841db0","model_key":"limite-1b-violetto-mlx","process_cpu_percent_avg":54.47089950027608,"process_rss_bytes_avg":96917494.20591466,"process_rss_bytes_peak":102842368.0,"run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","sample_count":670,"sample_interval_ms":331.7098112721101,"sampling_error_count":0,"scope":"server_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":7568949248.0},"run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","runtime_key":"korgis-local","sample_count":10,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":2837072896,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T15:45:29.838316+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b","execution_signature":"sha256:execution:0caf2fff221781464037ba78ff0ed5a8158d632ab0e66481fc8df733dd8785c6","experiment_kind":"standard","failure_count":0,"family":"nemotron-nano","generation_policy":{},"git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","parameters_b":4.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.5,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:0caf2fff221781464037ba78ff0ed5a8158d632ab0e66481fc8df733dd8785c6","model_key":"nemotron-nano-4b","process_cpu_percent_avg":10.014690415304136,"process_rss_bytes_avg":4240616618.6666665,"process_rss_bytes_peak":4889542656.0,"run_id":"3b0292d4-8f8f-4843-900c-946c32476964","sample_count":153,"sample_interval_ms":267.2538513759112,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5000953856.0},"run_id":"3b0292d4-8f8f-4843-900c-946c32476964","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":4233677952,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b-q8","execution_signature":"sha256:execution:89a74e35df26090fe2c6adb16b01e32373fbcdad6966a1b23cc9b88ba35a5170","experiment_kind":"standard","failure_count":0,"family":"nemotron-nano","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","parameters_b":4.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.475,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:89a74e35df26090fe2c6adb16b01e32373fbcdad6966a1b23cc9b88ba35a5170","model_key":"nemotron-nano-4b-q8","process_cpu_percent_avg":8.818030339465103,"process_rss_bytes_avg":7560277067.093333,"process_rss_bytes_peak":10620682240.0,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":152,"sample_interval_ms":260.26064477991895,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":10035691520.0},"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":532517120,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","execution_signature":"sha256:execution:c58f103ecf1090aab4510c6708cd9446fefc4137c61ae9ca33da8cc1c5d3faf8","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","parameters_b":0.8,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.1,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:c58f103ecf1090aab4510c6708cd9446fefc4137c61ae9ca33da8cc1c5d3faf8","model_key":"qwen3.5-0.8b-q4km","process_cpu_percent_avg":15.698688370479838,"process_rss_bytes_avg":1544970035.2,"process_rss_bytes_peak":2028224512.0,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":80,"sample_interval_ms":263.3856863947585,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5715886080.0},"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":3775709216,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-bf16","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","parameters_b":2.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.25,"profile":"core","provider_key":"korgis","quantization":"BF16","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","model_key":"qwen3.5-2b-bf16","process_cpu_percent_avg":8.718440372188164,"process_rss_bytes_avg":4446060373.333334,"process_rss_bytes_peak":4834885632.0,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":121,"sample_interval_ms":268.43950018325506,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5017370624.0},"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":1280835840,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T15:44:53.050810+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:8190c8f3bb1f7b51ef597577c913ceae05f8cb7998a7d5221e848f67071166c9","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","parameters_b":2.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.25,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:8190c8f3bb1f7b51ef597577c913ceae05f8cb7998a7d5221e848f67071166c9","model_key":"qwen3.5-2b-q4km","process_cpu_percent_avg":8.953256029208966,"process_rss_bytes_avg":2160430830.9333334,"process_rss_bytes_peak":2483290112.0,"run_id":"9fb9f932-1f29-4646-b337-f33476d841ba","sample_count":94,"sample_interval_ms":267.4467081844341,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5225758720.0},"run_id":"9fb9f932-1f29-4646-b337-f33476d841ba","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":2740937888,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:d6ef7d6b6d33f70d69c5d1df4568941686a610cda0bde994801472a902500dd4","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"unsloth/qwen3.5-4b","model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","parameters_b":4.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.4,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:d6ef7d6b6d33f70d69c5d1df4568941686a610cda0bde994801472a902500dd4","model_key":"qwen3.5-4b-q4km","process_cpu_percent_avg":10.67374362291546,"process_rss_bytes_avg":4568710143.999999,"process_rss_bytes_peak":5363400704.0,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":129,"sample_interval_ms":303.8685138652605,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5199380480.0},"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":4482403488,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q8","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","parameters_b":4.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.475,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","model_key":"qwen3.5-4b-q8","process_cpu_percent_avg":10.86561479591514,"process_rss_bytes_avg":5559348804.266666,"process_rss_bytes_peak":6319587328.0,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":146,"sample_interval_ms":295.3429861280407,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4841750528.0},"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":5627044256,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:9d728f687c9bb0f47a2c5a8c0e61b6da5a7e2833ad1afcffb18c7b4eb2767886","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","parameters_b":9.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.5,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:9d728f687c9bb0f47a2c5a8c0e61b6da5a7e2833ad1afcffb18c7b4eb2767886","model_key":"qwen3.5-9b-q4km","process_cpu_percent_avg":7.04412305796482,"process_rss_bytes_avg":8143566028.8,"process_rss_bytes_peak":9486303232.0,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":176,"sample_interval_ms":291.7408815034529,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":9481781248.0},"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":2600224352,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T15:45:57.052259+00:00","deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_signature":"sha256:execution:d26122294632fb1ef7e62445eece13d61030562e5a3d7a64b60606c3c12ff3be","experiment_kind":"standard","failure_count":0,"family":"spark-x2.5","generation_policy":{},"git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","model_id":"XHToken/Spark-X2.5-4B","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:8dcf43803761fd5570f637e166b655565b6c638e51a6a96848f58f6e6699a734","parameters_b":4.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.3,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:d26122294632fb1ef7e62445eece13d61030562e5a3d7a64b60606c3c12ff3be","model_key":"spark-x2.5-4b-q4km","process_cpu_percent_avg":6.338877896022156,"process_rss_bytes_avg":3725684701.8666673,"process_rss_bytes_peak":4221812736.0,"run_id":"e68754ab-25e6-4449-b06a-66dbc311d9e5","sample_count":122,"sample_interval_ms":266.60380743026815,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":8769241088.0},"run_id":"e68754ab-25e6-4449-b06a-66dbc311d9e5","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"}],"decision":{"capability_summaries":[{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:80688f79e2c27746cabb9c8e268a98d63ee3d732cb5c73d70fc581a1958017fc","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2179.467997944448,"latency_p50_ms":2134.854041505605,"latency_p95_ms":2559.561714925802,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","normalized_quality_score":47.5,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":357,"primary_metric":"final_answer_accuracy","primary_value":0.475,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":35,"failure_rate":0.875,"input_tokens_total":429,"known_provider_cost_quality_pareto":false,"latency_mean_ms":433.91745427506976,"latency_p50_ms":175.63410446746275,"latency_p95_ms":2067.160929192323,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","normalized_quality_score":12.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":34,"primary_metric":"final_answer_accuracy","primary_value":0.125,"provider_cost_coverage_rate":0.125,"provider_cost_known":true,"provider_cost_observed_total_usd":4.2375e-05,"provider_cost_per_1k_cases_usd":0.008475,"provider_cost_per_case_usd":8.475e-06,"provider_cost_priced_cases":5,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":0,"failure_rate":0.0,"input_tokens_total":4879,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1708.4744781750487,"latency_p50_ms":1610.1467295084149,"latency_p95_ms":2667.998251819517,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","normalized_quality_score":67.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":1724,"primary_metric":"final_answer_accuracy","primary_value":0.675,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0030446,"provider_cost_per_1k_cases_usd":0.076115,"provider_cost_per_case_usd":7.6115e-05,"provider_cost_priced_cases":40,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":40,"provider_cost_total_usd":0.0030446,"run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","sample_count":40},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":0,"failure_rate":0.0,"input_tokens_total":4879,"known_provider_cost_quality_pareto":true,"latency_mean_ms":1804.218880116241,"latency_p50_ms":1617.0925420010462,"latency_p95_ms":2760.3690872143484,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","normalized_quality_score":70.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":1670,"primary_metric":"final_answer_accuracy","primary_value":0.7,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0029798,"provider_cost_per_1k_cases_usd":0.07449499999999999,"provider_cost_per_case_usd":7.4495e-05,"provider_cost_priced_cases":40,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":40,"provider_cost_total_usd":0.0029798,"run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","sample_count":40},{"benchmark_signature":"sha256:consolidated:20f65d6394555078bc9dce0c1393e10d963aecb84ce9344ece12046a78d42de3","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","failure_count":0,"failure_rate":0.0,"input_tokens_total":4186,"known_provider_cost_quality_pareto":false,"latency_mean_ms":9138.54978744057,"latency_p50_ms":8435.180062428117,"latency_p95_ms":14727.237971138664,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","normalized_quality_score":95.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":20062,"primary_metric":"final_answer_accuracy","primary_value":0.95,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"lfm25-consolidated-latest","sample_count":40},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","failure_count":0,"failure_rate":0.0,"input_tokens_total":2437,"known_provider_cost_quality_pareto":false,"latency_mean_ms":14415.438993828138,"latency_p50_ms":12581.906624487601,"latency_p95_ms":29393.43815765458,"model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","normalized_quality_score":90.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":23911,"primary_metric":"final_answer_accuracy","primary_value":0.9,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"16a842b0-5dcc-47b5-8808-850b2ad1ac10","sample_count":40},{"benchmark_signature":"sha256:benchmark:dca92e0e83499d6053d9585c0bdee4321b7ec9651563312492a659578feb94ee","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:f4c681453ad106c8c878d601ee4a0ee6335674e81abaf3f7ba5297ce35841db0","failure_count":10,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":21810.495604295284,"latency_p50_ms":21736.863624944817,"latency_p95_ms":22921.96444603032,"model_key":"limite-1b-violetto-mlx","model_signature":"sha256:model:fddf01c0845257f40bbb635a92054df8eb22dacd450b827a951962e906d6513b","normalized_quality_score":0.0,"observed_case_count":10,"observed_quality_latency_pareto":false,"output_tokens_total":null,"primary_metric":"final_answer_accuracy","primary_value":0.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":10,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":10,"provider_cost_total_usd":null,"run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","sample_count":10},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:0caf2fff221781464037ba78ff0ed5a8158d632ab0e66481fc8df733dd8785c6","failure_count":0,"failure_rate":0.0,"input_tokens_total":4401,"known_provider_cost_quality_pareto":false,"latency_mean_ms":645.015773922205,"latency_p50_ms":618.5829375172034,"latency_p95_ms":876.0035978746599,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","normalized_quality_score":50.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":325,"primary_metric":"final_answer_accuracy","primary_value":0.5,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"3b0292d4-8f8f-4843-900c-946c32476964","sample_count":40},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:89a74e35df26090fe2c6adb16b01e32373fbcdad6966a1b23cc9b88ba35a5170","failure_count":0,"failure_rate":0.0,"input_tokens_total":4401,"known_provider_cost_quality_pareto":false,"latency_mean_ms":630.3394749935251,"latency_p50_ms":605.8572710026056,"latency_p95_ms":712.8466440772164,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","normalized_quality_score":47.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":335,"primary_metric":"final_answer_accuracy","primary_value":0.475,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:c58f103ecf1090aab4510c6708cd9446fefc4137c61ae9ca33da8cc1c5d3faf8","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":159.1911823285045,"latency_p50_ms":154.29127059178427,"latency_p95_ms":192.66062675160353,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","normalized_quality_score":10.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":345,"primary_metric":"final_answer_accuracy","primary_value":0.1,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":423.17552603199147,"latency_p50_ms":399.95974994963035,"latency_p95_ms":596.0868246445898,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","normalized_quality_score":25.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":351,"primary_metric":"final_answer_accuracy","primary_value":0.25,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:8190c8f3bb1f7b51ef597577c913ceae05f8cb7998a7d5221e848f67071166c9","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":264.1106479219161,"latency_p50_ms":243.21977101499215,"latency_p95_ms":369.4675250735594,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","normalized_quality_score":25.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":343,"primary_metric":"final_answer_accuracy","primary_value":0.25,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"9fb9f932-1f29-4646-b337-f33476d841ba","sample_count":40},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:d6ef7d6b6d33f70d69c5d1df4568941686a610cda0bde994801472a902500dd4","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":550.896484364057,"latency_p50_ms":541.7198544600978,"latency_p95_ms":659.4784934190102,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","normalized_quality_score":40.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":324,"primary_metric":"final_answer_accuracy","primary_value":0.4,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":638.2667771249544,"latency_p50_ms":647.2247504862025,"latency_p95_ms":751.9698580726981,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","normalized_quality_score":47.5,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":325,"primary_metric":"final_answer_accuracy","primary_value":0.475,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:9d728f687c9bb0f47a2c5a8c0e61b6da5a7e2833ad1afcffb18c7b4eb2767886","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":890.66908233508,"latency_p50_ms":886.8744795327075,"latency_p95_ms":1010.3516017552464,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","normalized_quality_score":50.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":327,"primary_metric":"final_answer_accuracy","primary_value":0.5,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:d26122294632fb1ef7e62445eece13d61030562e5a3d7a64b60606c3c12ff3be","failure_count":0,"failure_rate":0.0,"input_tokens_total":4493,"known_provider_cost_quality_pareto":false,"latency_mean_ms":483.35021561942995,"latency_p50_ms":480.80327099887654,"latency_p95_ms":562.0952548051717,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:8dcf43803761fd5570f637e166b655565b6c638e51a6a96848f58f6e6699a734","normalized_quality_score":30.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":346,"primary_metric":"final_answer_accuracy","primary_value":0.3,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"e68754ab-25e6-4449-b06a-66dbc311d9e5","sample_count":40}],"dataset_summaries":[{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2179.467997944448,"latency_p50_ms":2134.854041505605,"latency_p95_ms":2559.561714925802,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","normalized_quality_score":47.5,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":357,"primary_metric":"final_answer_accuracy","primary_value":0.475,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":35,"failure_rate":0.875,"input_tokens_total":429,"known_provider_cost_quality_pareto":false,"latency_mean_ms":433.91745427506976,"latency_p50_ms":175.63410446746275,"latency_p95_ms":2067.160929192323,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","normalized_quality_score":12.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":34,"primary_metric":"final_answer_accuracy","primary_value":0.125,"provider_cost_coverage_rate":0.125,"provider_cost_known":true,"provider_cost_observed_total_usd":4.2375e-05,"provider_cost_per_1k_cases_usd":0.008475,"provider_cost_per_case_usd":8.475e-06,"provider_cost_priced_cases":5,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4879,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1708.4744781750487,"latency_p50_ms":1610.1467295084149,"latency_p95_ms":2667.998251819517,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","normalized_quality_score":67.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":1724,"primary_metric":"final_answer_accuracy","primary_value":0.675,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0030446,"provider_cost_per_1k_cases_usd":0.076115,"provider_cost_per_case_usd":7.6115e-05,"provider_cost_priced_cases":40,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":40,"provider_cost_total_usd":0.0030446,"sample_count":40},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4879,"known_provider_cost_quality_pareto":true,"latency_mean_ms":1804.218880116241,"latency_p50_ms":1617.0925420010462,"latency_p95_ms":2760.3690872143484,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","normalized_quality_score":70.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":1670,"primary_metric":"final_answer_accuracy","primary_value":0.7,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0029798,"provider_cost_per_1k_cases_usd":0.07449499999999999,"provider_cost_per_case_usd":7.4495e-05,"provider_cost_priced_cases":40,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":40,"provider_cost_total_usd":0.0029798,"sample_count":40},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4186,"known_provider_cost_quality_pareto":false,"latency_mean_ms":9138.54978744057,"latency_p50_ms":8435.180062428117,"latency_p95_ms":14727.237971138664,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","normalized_quality_score":95.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":20062,"primary_metric":"final_answer_accuracy","primary_value":0.95,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":2437,"known_provider_cost_quality_pareto":false,"latency_mean_ms":14415.438993828138,"latency_p50_ms":12581.906624487601,"latency_p95_ms":29393.43815765458,"model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","normalized_quality_score":90.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":23911,"primary_metric":"final_answer_accuracy","primary_value":0.9,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":10,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":21810.495604295284,"latency_p50_ms":21736.863624944817,"latency_p95_ms":22921.96444603032,"model_key":"limite-1b-violetto-mlx","model_signature":"sha256:model:fddf01c0845257f40bbb635a92054df8eb22dacd450b827a951962e906d6513b","normalized_quality_score":0.0,"observed_case_count":10,"observed_quality_latency_pareto":false,"output_tokens_total":null,"primary_metric":"final_answer_accuracy","primary_value":0.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":10,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":10,"provider_cost_total_usd":null,"sample_count":10},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4401,"known_provider_cost_quality_pareto":false,"latency_mean_ms":645.015773922205,"latency_p50_ms":618.5829375172034,"latency_p95_ms":876.0035978746599,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","normalized_quality_score":50.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":325,"primary_metric":"final_answer_accuracy","primary_value":0.5,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4401,"known_provider_cost_quality_pareto":false,"latency_mean_ms":630.3394749935251,"latency_p50_ms":605.8572710026056,"latency_p95_ms":712.8466440772164,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","normalized_quality_score":47.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":335,"primary_metric":"final_answer_accuracy","primary_value":0.475,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":159.1911823285045,"latency_p50_ms":154.29127059178427,"latency_p95_ms":192.66062675160353,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","normalized_quality_score":10.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":345,"primary_metric":"final_answer_accuracy","primary_value":0.1,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":423.17552603199147,"latency_p50_ms":399.95974994963035,"latency_p95_ms":596.0868246445898,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","normalized_quality_score":25.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":351,"primary_metric":"final_answer_accuracy","primary_value":0.25,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":264.1106479219161,"latency_p50_ms":243.21977101499215,"latency_p95_ms":369.4675250735594,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","normalized_quality_score":25.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":343,"primary_metric":"final_answer_accuracy","primary_value":0.25,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":550.896484364057,"latency_p50_ms":541.7198544600978,"latency_p95_ms":659.4784934190102,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","normalized_quality_score":40.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":324,"primary_metric":"final_answer_accuracy","primary_value":0.4,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":638.2667771249544,"latency_p50_ms":647.2247504862025,"latency_p95_ms":751.9698580726981,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","normalized_quality_score":47.5,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":325,"primary_metric":"final_answer_accuracy","primary_value":0.475,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":890.66908233508,"latency_p50_ms":886.8744795327075,"latency_p95_ms":1010.3516017552464,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","normalized_quality_score":50.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":327,"primary_metric":"final_answer_accuracy","primary_value":0.5,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4493,"known_provider_cost_quality_pareto":false,"latency_mean_ms":483.35021561942995,"latency_p50_ms":480.80327099887654,"latency_p95_ms":562.0952548051717,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:8dcf43803761fd5570f637e166b655565b6c638e51a6a96848f58f6e6699a734","normalized_quality_score":30.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":346,"primary_metric":"final_answer_accuracy","primary_value":0.3,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40}]},"disagreements":[],"family_breakdown":[],"schema_version":"2"},"mathematical-reasoning-v2":{"benchmark_signatures":["sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","sha256:benchmark:600e8b1d94c1e2bc55bd8802918ce0c6f18dece6d8a9c02247a530fc7b015409","sha256:benchmark:bff2038c9a70a239e320793c0c579d02b35645dd0f722e6fd8e0c6a9b30b4b07"],"capability_id":"mathematical-reasoning-v2","cases":[],"cells":[{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-06T09:40:29.476460+00:00","deployment":"local","effective_model_id":"bonsai-27b-q10","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.575,"profile":"core","provider_key":"korgis","quantization":"Q1_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"bonsai-27b-q10","process_cpu_percent_avg":10.398893667073832,"process_rss_bytes_avg":7636308908.292063,"process_rss_bytes_peak":11685068800.0,"run_id":"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","sample_count":340,"sample_interval_ms":307.1656581635226,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5910167552.0},"run_id":"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T13:59:36.375684+00:00","deployment":"api","effective_model_id":"gemini-3.5-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.5","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.0,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core","runtime_key":"gemini-api","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T13:38:58.640898+00:00","deployment":"api","effective_model_id":"gemini-3.8-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.8","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.8-flash","model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.0,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core","runtime_key":"gemini-api","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-06T12:17:36.689339+00:00","deployment":"api","effective_model_id":"gpt-5.6-luna","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.45,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core","runtime_key":"openai-api","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T13:13:20.819089+00:00","deployment":"api","effective_model_id":"gpt-6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","experiment_kind":"standard","failure_count":0,"family":"gpt-6","generation_policy":{},"git_commit":"7f796c10d99e934da6417ea54405fbf9352ad399","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.425,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core","runtime_key":"openai-api","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":10947100928,"benchmark_signature":"sha256:benchmark:600e8b1d94c1e2bc55bd8802918ce0c6f18dece6d8a9c02247a530fc7b015409","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-22T14:07:00.510877+00:00","deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_signature":"sha256:execution:503d302009e21bb18588b87ac3897f9112994a479905bbca872384efa4360298","experiment_kind":"standard","failure_count":0,"family":"lfm2.5","generation_policy":{"max_output_tokens":4096},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.775,"profile":"core","provider_key":"korgis","quantization":"UD-Q3_K_XL","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:503d302009e21bb18588b87ac3897f9112994a479905bbca872384efa4360298","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","process_cpu_percent_avg":10.29314878109045,"process_rss_bytes_avg":1417688467.7424607,"process_rss_bytes_peak":1463648256.0,"run_id":"lfm25-math-v2-core","sample_count":1316,"sample_interval_ms":313.23082923487743,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4090920960.0},"run_id":"lfm25-math-v2-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-17T21:24:56.752644+00:00","deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","experiment_kind":"standard","failure_count":0,"family":"limite","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","parameters_b":1.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.8,"profile":"core","provider_key":"transformers","quantization":null,"resource_summary":null,"run_id":"8eef4bab-617a-4638-9e51-53e72034e821","runtime_key":"transformers-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:bff2038c9a70a239e320793c0c579d02b35645dd0f722e6fd8e0c6a9b30b4b07","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T13:35:21.786434+00:00","deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","experiment_kind":"standard","failure_count":0,"family":"limite","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65","parameters_b":1.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.8333333333333334,"profile":"smoke","provider_key":"transformers","quantization":null,"resource_summary":null,"run_id":"cb837fcf-570f-4a89-bbbd-cded7a3c335b","runtime_key":"transformers-local","sample_count":12,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T12:48:39.598492+00:00","deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:ff14676ac39e23af360cd1209491ef1c14e3dab9a3198b48054de5c6de699fdb","experiment_kind":"standard","failure_count":0,"family":"limite","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:3fe1bfff5bb2c952f96bb799970eb76d9db17f9f880bc3326f0bdfbdd51577cc","parameters_b":1.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.0,"profile":"core","provider_key":"vllm","quantization":null,"resource_summary":null,"run_id":"5623e5c1-213d-447a-88e9-ac51a69a92fa","runtime_key":"vllm-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T13:46:02.371339+00:00","deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","experiment_kind":"standard","failure_count":0,"family":"limite","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65","parameters_b":1.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.725,"profile":"core","provider_key":"transformers","quantization":null,"resource_summary":null,"run_id":"954053b1-6f87-4196-bb2a-028404b6fef1","runtime_key":"transformers-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T11:57:43.663487+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.575,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b","process_cpu_percent_avg":11.038100201393934,"process_rss_bytes_avg":7133634082.133334,"process_rss_bytes_peak":11070996480.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core","sample_count":143,"sample_interval_ms":255.84691734608595,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":11042144256.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T13:15:22.380707+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b-q8","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.6,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b-q8","process_cpu_percent_avg":8.601215146103618,"process_rss_bytes_avg":7892413781.333334,"process_rss_bytes_peak":11438456832.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","sample_count":158,"sample_interval_ms":252.1124605016667,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":7386234880.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T13:36:13.768283+00:00","deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.2,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-0.8b-q4km","process_cpu_percent_avg":18.636242311523596,"process_rss_bytes_avg":1745210777.6,"process_rss_bytes_peak":2670067712.0,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core","sample_count":81,"sample_interval_ms":213.9264905885284,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":15259779072.0},"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":3775709216,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T11:22:33.733080+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-bf16","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","parameters_b":2.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.375,"profile":"core","provider_key":"korgis","quantization":"BF16","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","model_key":"qwen3.5-2b-bf16","process_cpu_percent_avg":6.761276471009742,"process_rss_bytes_avg":5075235362.133334,"process_rss_bytes_peak":5976489984.0,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core","sample_count":120,"sample_interval_ms":240.70379737386247,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":10141270016.0},"run_id":"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T11:23:17.557837+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.325,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-2b-q4km","process_cpu_percent_avg":12.866597281256125,"process_rss_bytes_avg":2496888695.466666,"process_rss_bytes_peak":3418226688.0,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core","sample_count":104,"sample_interval_ms":243.8169525759804,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":8507686912.0},"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T16:36:39.120802+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"cb061291446b8dbd79f524a4fefcd4a4406e2a0d","model_id":"unsloth/qwen3.5-4b","model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.475,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-4b-q4km","process_cpu_percent_avg":12.689180073192913,"process_rss_bytes_avg":5970077900.800001,"process_rss_bytes_peak":7198195712.0,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":134,"sample_interval_ms":272.2716644006747,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":8184053760.0},"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":4482403488,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T11:32:22.918386+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q8","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","parameters_b":4.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.5,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","model_key":"qwen3.5-4b-q8","process_cpu_percent_avg":6.982037738154942,"process_rss_bytes_avg":7516327697.066668,"process_rss_bytes_peak":9962160128.0,"run_id":"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core","sample_count":132,"sample_interval_ms":291.2576532937237,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":11326062592.0},"run_id":"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T12:39:03.216607+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.675,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-9b-q4km","process_cpu_percent_avg":6.526783901766963,"process_rss_bytes_avg":8611465482.24,"process_rss_bytes_peak":11088855040.0,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core","sample_count":175,"sample_interval_ms":275.92921112542774,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":11636277248.0},"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":4121781472,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T11:48:03.359698+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-ud-q2kxl","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"qwen/qwen3.5-9b-ud-q2kxl","model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","parameters_b":9.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.55,"profile":"core","provider_key":"korgis","quantization":"UD-Q2_K_XL","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","model_key":"qwen3.5-9b-ud-q2kxl","process_cpu_percent_avg":4.913551711729864,"process_rss_bytes_avg":7164752896.0,"process_rss_bytes_peak":9602973696.0,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core","sample_count":164,"sample_interval_ms":288.91353003637056,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":12627738624.0},"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":6192222208,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-06T14:37:12.067024+00:00","deployment":"local","effective_model_id":"qwen3.8-27b-iq1s","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","experiment_kind":"standard","failure_count":0,"family":"qwen3.8","generation_policy":{},"git_commit":"640138c3429882a42797d28074f98d3fea98a63b","model_id":"Qwen/Qwen3.8-27B","model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","parameters_b":27.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.25,"profile":"core","provider_key":"korgis","quantization":"UD-IQ1_S","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","model_key":"qwen3.8-27b-iq1s","process_cpu_percent_avg":7.113533739353301,"process_rss_bytes_avg":9138898947.755379,"process_rss_bytes_peak":12174065664.0,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","sample_count":717,"sample_interval_ms":297.72640370809415,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5485953024.0},"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-06T09:01:45.488881+00:00","deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"xhtoken/spark-x2.5-4b","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.45,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"spark-x2.5-4b-q4km","process_cpu_percent_avg":8.149214468504207,"process_rss_bytes_avg":3498299289.6,"process_rss_bytes_peak":3887923200.0,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core","sample_count":147,"sample_interval_ms":267.7047739123735,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":7349649408.0},"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"}],"decision":{"capability_summaries":[{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2219.316747926132,"latency_p50_ms":2213.482687504438,"latency_p95_ms":2470.525904195528,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","normalized_quality_score":57.49999999999999,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":396,"primary_metric":"final_answer_accuracy","primary_value":0.575,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":40,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":134.92541666782927,"latency_p50_ms":129.18631246429868,"latency_p95_ms":178.56887083326,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","normalized_quality_score":0.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":null,"primary_metric":"final_answer_accuracy","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":40,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":135.2158302499447,"latency_p50_ms":129.64654198731296,"latency_p95_ms":177.9962440428789,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","normalized_quality_score":0.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":null,"primary_metric":"final_answer_accuracy","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.8-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4745,"known_provider_cost_quality_pareto":true,"latency_mean_ms":1729.619964624726,"latency_p50_ms":1645.4874375049258,"latency_p95_ms":2669.140464547672,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","normalized_quality_score":45.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":1415,"primary_metric":"final_answer_accuracy","primary_value":0.45,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.002647,"provider_cost_per_1k_cases_usd":0.066175,"provider_cost_per_case_usd":6.6175e-05,"provider_cost_priced_cases":40,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":40,"provider_cost_total_usd":0.002647,"run_id":"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":0,"failure_rate":0.0,"input_tokens_total":4745,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1796.6679405944888,"latency_p50_ms":1630.790083494503,"latency_p95_ms":2689.1042798262743,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","normalized_quality_score":42.5,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":1577,"primary_metric":"final_answer_accuracy","primary_value":0.425,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0028414,"provider_cost_per_1k_cases_usd":0.071035,"provider_cost_per_case_usd":7.1035e-05,"provider_cost_priced_cases":40,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":40,"provider_cost_total_usd":0.0028414,"run_id":"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:600e8b1d94c1e2bc55bd8802918ce0c6f18dece6d8a9c02247a530fc7b015409","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:503d302009e21bb18588b87ac3897f9112994a479905bbca872384efa4360298","failure_count":0,"failure_rate":0.0,"input_tokens_total":4068,"known_provider_cost_quality_pareto":false,"latency_mean_ms":9981.023337302031,"latency_p50_ms":8236.826166510582,"latency_p95_ms":23799.921930843244,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","normalized_quality_score":77.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":22004,"primary_metric":"final_answer_accuracy","primary_value":0.775,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"lfm25-math-v2-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","failure_count":2,"failure_rate":0.05,"input_tokens_total":2323,"known_provider_cost_quality_pareto":false,"latency_mean_ms":16638.17807911255,"latency_p50_ms":12906.520646007266,"latency_p95_ms":49143.83464824058,"model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","normalized_quality_score":80.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":27084,"primary_metric":"final_answer_accuracy","primary_value":0.8,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"8eef4bab-617a-4638-9e51-53e72034e821","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":4295,"known_provider_cost_quality_pareto":false,"latency_mean_ms":574.8850365260296,"latency_p50_ms":566.8227080022916,"latency_p95_ms":664.7058172944526,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","normalized_quality_score":57.49999999999999,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":326,"primary_metric":"final_answer_accuracy","primary_value":0.575,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":4295,"known_provider_cost_quality_pareto":false,"latency_mean_ms":682.1980853499554,"latency_p50_ms":653.759083004843,"latency_p95_ms":789.8509145503339,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","normalized_quality_score":60.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":326,"primary_metric":"final_answer_accuracy","primary_value":0.6,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":162.14768229983747,"latency_p50_ms":147.21993750572437,"latency_p95_ms":202.90230384853203,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","normalized_quality_score":20.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":410,"primary_metric":"final_answer_accuracy","primary_value":0.2,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":402.97180729248794,"latency_p50_ms":390.3209579875693,"latency_p95_ms":501.28230001137115,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","normalized_quality_score":37.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":371,"primary_metric":"final_answer_accuracy","primary_value":0.375,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":266.83754889891134,"latency_p50_ms":261.64224999956787,"latency_p95_ms":332.65334790921764,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","normalized_quality_score":32.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":371,"primary_metric":"final_answer_accuracy","primary_value":0.325,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":557.498613624557,"latency_p50_ms":543.7640624950291,"latency_p95_ms":709.5650934068543,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","normalized_quality_score":47.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":328,"primary_metric":"final_answer_accuracy","primary_value":0.475,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":571.6949749272317,"latency_p50_ms":558.6196244985331,"latency_p95_ms":686.1018520867218,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","normalized_quality_score":50.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":328,"primary_metric":"final_answer_accuracy","primary_value":0.5,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":857.8194458481448,"latency_p50_ms":846.0788744923775,"latency_p95_ms":999.0503857392467,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","normalized_quality_score":67.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":333,"primary_metric":"final_answer_accuracy","primary_value":0.675,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":817.9967738789856,"latency_p50_ms":805.0282710173633,"latency_p95_ms":935.7486851222346,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","normalized_quality_score":55.00000000000001,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":329,"primary_metric":"final_answer_accuracy","primary_value":0.55,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","failure_count":7,"failure_rate":0.175,"input_tokens_total":4075,"known_provider_cost_quality_pareto":false,"latency_mean_ms":5066.640841649496,"latency_p50_ms":2834.1644589963835,"latency_p95_ms":22229.72979789554,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","normalized_quality_score":25.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":719,"primary_metric":"final_answer_accuracy","primary_value":0.25,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":4373,"known_provider_cost_quality_pareto":false,"latency_mean_ms":619.3483061986626,"latency_p50_ms":629.36987500143,"latency_p95_ms":752.714208359248,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","normalized_quality_score":45.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":346,"primary_metric":"final_answer_accuracy","primary_value":0.45,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core","sample_count":40}],"dataset_summaries":[{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2219.316747926132,"latency_p50_ms":2213.482687504438,"latency_p95_ms":2470.525904195528,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","normalized_quality_score":57.49999999999999,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":396,"primary_metric":"final_answer_accuracy","primary_value":0.575,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":40,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":134.92541666782927,"latency_p50_ms":129.18631246429868,"latency_p95_ms":178.56887083326,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","normalized_quality_score":0.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":null,"primary_metric":"final_answer_accuracy","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":40,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":135.2158302499447,"latency_p50_ms":129.64654198731296,"latency_p95_ms":177.9962440428789,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","normalized_quality_score":0.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":null,"primary_metric":"final_answer_accuracy","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.8-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4745,"known_provider_cost_quality_pareto":true,"latency_mean_ms":1729.619964624726,"latency_p50_ms":1645.4874375049258,"latency_p95_ms":2669.140464547672,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","normalized_quality_score":45.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":1415,"primary_metric":"final_answer_accuracy","primary_value":0.45,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.002647,"provider_cost_per_1k_cases_usd":0.066175,"provider_cost_per_case_usd":6.6175e-05,"provider_cost_priced_cases":40,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":40,"provider_cost_total_usd":0.002647,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4745,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1796.6679405944888,"latency_p50_ms":1630.790083494503,"latency_p95_ms":2689.1042798262743,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","normalized_quality_score":42.5,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":1577,"primary_metric":"final_answer_accuracy","primary_value":0.425,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0028414,"provider_cost_per_1k_cases_usd":0.071035,"provider_cost_per_case_usd":7.1035e-05,"provider_cost_priced_cases":40,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":40,"provider_cost_total_usd":0.0028414,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4068,"known_provider_cost_quality_pareto":false,"latency_mean_ms":9981.023337302031,"latency_p50_ms":8236.826166510582,"latency_p95_ms":23799.921930843244,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","normalized_quality_score":77.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":22004,"primary_metric":"final_answer_accuracy","primary_value":0.775,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":2,"failure_rate":0.05,"input_tokens_total":2323,"known_provider_cost_quality_pareto":false,"latency_mean_ms":16638.17807911255,"latency_p50_ms":12906.520646007266,"latency_p95_ms":49143.83464824058,"model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","normalized_quality_score":80.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":27084,"primary_metric":"final_answer_accuracy","primary_value":0.8,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4295,"known_provider_cost_quality_pareto":false,"latency_mean_ms":574.8850365260296,"latency_p50_ms":566.8227080022916,"latency_p95_ms":664.7058172944526,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","normalized_quality_score":57.49999999999999,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":326,"primary_metric":"final_answer_accuracy","primary_value":0.575,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4295,"known_provider_cost_quality_pareto":false,"latency_mean_ms":682.1980853499554,"latency_p50_ms":653.759083004843,"latency_p95_ms":789.8509145503339,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","normalized_quality_score":60.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":326,"primary_metric":"final_answer_accuracy","primary_value":0.6,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":162.14768229983747,"latency_p50_ms":147.21993750572437,"latency_p95_ms":202.90230384853203,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","normalized_quality_score":20.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":410,"primary_metric":"final_answer_accuracy","primary_value":0.2,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":402.97180729248794,"latency_p50_ms":390.3209579875693,"latency_p95_ms":501.28230001137115,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","normalized_quality_score":37.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":371,"primary_metric":"final_answer_accuracy","primary_value":0.375,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":266.83754889891134,"latency_p50_ms":261.64224999956787,"latency_p95_ms":332.65334790921764,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","normalized_quality_score":32.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":371,"primary_metric":"final_answer_accuracy","primary_value":0.325,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":557.498613624557,"latency_p50_ms":543.7640624950291,"latency_p95_ms":709.5650934068543,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","normalized_quality_score":47.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":328,"primary_metric":"final_answer_accuracy","primary_value":0.475,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":571.6949749272317,"latency_p50_ms":558.6196244985331,"latency_p95_ms":686.1018520867218,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","normalized_quality_score":50.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":328,"primary_metric":"final_answer_accuracy","primary_value":0.5,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":857.8194458481448,"latency_p50_ms":846.0788744923775,"latency_p95_ms":999.0503857392467,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","normalized_quality_score":67.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":333,"primary_metric":"final_answer_accuracy","primary_value":0.675,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":817.9967738789856,"latency_p50_ms":805.0282710173633,"latency_p95_ms":935.7486851222346,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","normalized_quality_score":55.00000000000001,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":329,"primary_metric":"final_answer_accuracy","primary_value":0.55,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":7,"failure_rate":0.175,"input_tokens_total":4075,"known_provider_cost_quality_pareto":false,"latency_mean_ms":5066.640841649496,"latency_p50_ms":2834.1644589963835,"latency_p95_ms":22229.72979789554,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","normalized_quality_score":25.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":719,"primary_metric":"final_answer_accuracy","primary_value":0.25,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4373,"known_provider_cost_quality_pareto":false,"latency_mean_ms":619.3483061986626,"latency_p50_ms":629.36987500143,"latency_p95_ms":752.714208359248,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","normalized_quality_score":45.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":346,"primary_metric":"final_answer_accuracy","primary_value":0.45,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40}]},"disagreements":[],"family_breakdown":[],"schema_version":"2"},"oos-calibration":{"benchmark_signatures":["sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","sha256:consolidated:9468d75aebe9e805cefc3643f61a9e5fe6127c7a4e3a987608b30841b9bc9fa0"],"capability_id":"oos-calibration","cases":[],"cells":[{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-06T10:26:04.512560+00:00","deployment":"local","effective_model_id":"bonsai-27b-q10","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.948051948051948,"profile":"core","provider_key":"korgis","quantization":"Q1_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"bonsai-27b-q10","process_cpu_percent_avg":3.871693631322875,"process_rss_bytes_avg":5867857169.246841,"process_rss_bytes_peak":9145516032.0,"run_id":"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core","sample_count":4524,"sample_interval_ms":309.6376288065501,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5604933632.0},"run_id":"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T14:00:21.676846+00:00","deployment":"api","effective_model_id":"gemini-3.5-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.5","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.5,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core","runtime_key":"gemini-api","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T13:39:43.452981+00:00","deployment":"api","effective_model_id":"gemini-3.8-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.8","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.8-flash","model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.5,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core","runtime_key":"gemini-api","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-06T12:26:04.935243+00:00","deployment":"api","effective_model_id":"gpt-5.6-luna","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"11b00f55c776ec89ddee7cd2dd1898797dde4055","model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":1.0,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core","runtime_key":"openai-api","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T13:25:20.495814+00:00","deployment":"api","effective_model_id":"gpt-6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","experiment_kind":"standard","failure_count":0,"family":"gpt-6","generation_policy":{},"git_commit":"7f796c10d99e934da6417ea54405fbf9352ad399","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.9935064935064936,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"gpt-6-luna-core-gpt-6-luna-oos-calibration-core","runtime_key":"openai-api","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-06T12:51:50.754874+00:00","deployment":"api","effective_model_id":"jev-1.13.0","execution_signature":"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3","experiment_kind":"standard","failure_count":0,"family":"jev","generation_policy":{},"git_commit":"72697127df45f8a3abc98b0b7bdf6fd53a0b5c98","model_id":"jev-1.13.0","model_key":"jev-1.13.0","model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.974025974025974,"profile":"core","provider_key":"typesafe","quantization":null,"resource_summary":null,"run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","runtime_key":"typesafe-api","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":10947100928,"benchmark_signature":"sha256:consolidated:9468d75aebe9e805cefc3643f61a9e5fe6127c7a4e3a987608b30841b9bc9fa0","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-22T13:12:52.242438+00:00","deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","experiment_kind":"consolidated","failure_count":0,"family":"lfm2.5","generation_policy":{"generation_configs":[{"extra":{},"max_output_tokens":2048,"seed":42,"stop":[],"temperature":0.0},{"extra":{},"max_output_tokens":4096,"seed":42,"stop":[],"temperature":0.0}],"mixed_generation":true},"git_commit":null,"model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.9545454545454546,"profile":"core","provider_key":"korgis","quantization":"UD-Q3_K_XL","resource_summary":null,"run_id":"lfm25-consolidated-latest","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T12:10:00.371141+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b","process_cpu_percent_avg":5.490567532375733,"process_rss_bytes_avg":5130473995.082253,"process_rss_bytes_peak":10141679616.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core","sample_count":1444,"sample_interval_ms":296.66646442619066,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6067978240.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T13:32:50.472664+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b-q8","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b-q8","process_cpu_percent_avg":5.151403915249351,"process_rss_bytes_avg":5654897500.709869,"process_rss_bytes_peak":8370143232.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core","sample_count":1747,"sample_interval_ms":306.91440564565875,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5491818496.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T13:39:04.217888+00:00","deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.5,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-0.8b-q4km","process_cpu_percent_avg":16.11579774911694,"process_rss_bytes_avg":5182799526.2337675,"process_rss_bytes_peak":9496838144.0,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","sample_count":463,"sample_interval_ms":247.8098196266941,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":9132261376.0},"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":3775709216,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T11:28:16.241563+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-bf16","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","parameters_b":2.0,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.5,"profile":"core","provider_key":"korgis","quantization":"BF16","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","model_key":"qwen3.5-2b-bf16","process_cpu_percent_avg":5.117048169336818,"process_rss_bytes_avg":5994336968.810392,"process_rss_bytes_peak":9291366400.0,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","sample_count":770,"sample_interval_ms":293.5350881957497,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5950177280.0},"run_id":"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T11:28:16.098461+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.5,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-2b-q4km","process_cpu_percent_avg":8.346543248013683,"process_rss_bytes_avg":3164761335.3558445,"process_rss_bytes_peak":6107152384.0,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","sample_count":621,"sample_interval_ms":291.97748509630554,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6536364032.0},"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T16:36:39.120802+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"cb061291446b8dbd79f524a4fefcd4a4406e2a0d","model_id":"unsloth/qwen3.5-4b","model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.974025974025974,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-4b-q4km","process_cpu_percent_avg":6.406836413976952,"process_rss_bytes_avg":6021487127.958771,"process_rss_bytes_peak":10610851840.0,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":1193,"sample_interval_ms":286.7970046134617,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5619220480.0},"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":4482403488,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T11:43:07.482305+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q8","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","parameters_b":4.0,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.9805194805194806,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","model_key":"qwen3.5-4b-q8","process_cpu_percent_avg":5.610485336009209,"process_rss_bytes_avg":7096272805.442177,"process_rss_bytes_peak":11637391360.0,"run_id":"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","sample_count":1211,"sample_interval_ms":319.76690310093494,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5589811200.0},"run_id":"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T13:01:17.635848+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.974025974025974,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-9b-q4km","process_cpu_percent_avg":2.683081412939199,"process_rss_bytes_avg":6844874999.319381,"process_rss_bytes_peak":10287939584.0,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core","sample_count":2848,"sample_interval_ms":314.6654996565356,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5142609920.0},"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":4121781472,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T12:04:10.873473+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-ud-q2kxl","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"qwen/qwen3.5-9b-ud-q2kxl","model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","parameters_b":9.0,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"profile":"core","provider_key":"korgis","quantization":"UD-Q2_K_XL","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","model_key":"qwen3.5-9b-ud-q2kxl","process_cpu_percent_avg":3.489486960210178,"process_rss_bytes_avg":6127369651.484418,"process_rss_bytes_peak":9636659200.0,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core","sample_count":1713,"sample_interval_ms":322.34991046526653,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5265096704.0},"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":6192222208,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-06T14:53:01.287222+00:00","deployment":"local","effective_model_id":"qwen3.8-27b-iq1s","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","experiment_kind":"standard","failure_count":0,"family":"qwen3.8","generation_policy":{},"git_commit":"640138c3429882a42797d28074f98d3fea98a63b","model_id":"Qwen/Qwen3.8-27B","model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","parameters_b":27.0,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.8181818181818182,"profile":"core","provider_key":"korgis","quantization":"UD-IQ1_S","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","model_key":"qwen3.8-27b-iq1s","process_cpu_percent_avg":3.3561674118938543,"process_rss_bytes_avg":8080732806.054676,"process_rss_bytes_peak":11618025472.0,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","sample_count":6492,"sample_interval_ms":307.9710546548535,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5336121344.0},"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-06T09:14:39.475593+00:00","deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"xhtoken/spark-x2.5-4b","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"spark-x2.5-4b-q4km","process_cpu_percent_avg":5.353823389553569,"process_rss_bytes_avg":4696997069.724959,"process_rss_bytes_peak":7862026240.0,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","sample_count":1416,"sample_interval_ms":302.2578173696654,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6040551424.0},"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"}],"decision":{"capability_summaries":[{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":93984,"known_provider_cost_quality_pareto":false,"latency_mean_ms":8717.480386934201,"latency_p50_ms":8677.232062495023,"latency_p95_ms":9162.466885308095,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","normalized_quality_score":94.8051948051948,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":3634,"primary_metric":"oos_detection_rate","primary_value":0.948051948051948,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":154,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":137.23104790891762,"latency_p50_ms":126.3813539990224,"latency_p95_ms":179.44677726482038,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","normalized_quality_score":50.0,"observed_case_count":154,"observed_quality_latency_pareto":true,"output_tokens_total":null,"primary_metric":"oos_detection_rate","primary_value":0.5,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":154,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":130.86163958908034,"latency_p50_ms":128.1952289864421,"latency_p95_ms":157.77898573724084,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","normalized_quality_score":50.0,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":null,"primary_metric":"oos_detection_rate","primary_value":0.5,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.8-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","failure_count":4,"failure_rate":0.025974025974025976,"input_tokens_total":157214,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1541.9887418898584,"latency_p50_ms":1287.3492500002612,"latency_p95_ms":2838.6875687552674,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","normalized_quality_score":100.0,"observed_case_count":154,"observed_quality_latency_pareto":true,"output_tokens_total":5434,"primary_metric":"oos_detection_rate","primary_value":1.0,"provider_cost_coverage_rate":0.974025974025974,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0379636,"provider_cost_per_1k_cases_usd":0.2530906666666667,"provider_cost_per_case_usd":0.00025309066666666667,"provider_cost_priced_cases":150,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":1,"failure_rate":0.006493506493506494,"input_tokens_total":160355,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2620.117901461184,"latency_p50_ms":2474.8345834668726,"latency_p95_ms":4364.541489130352,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","normalized_quality_score":99.35064935064936,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":8856,"primary_metric":"oos_detection_rate","primary_value":0.9935064935064936,"provider_cost_coverage_rate":0.9935064935064936,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0426982,"provider_cost_per_1k_cases_usd":0.27907320261437907,"provider_cost_per_case_usd":0.00027907320261437907,"provider_cost_priced_cases":153,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"gpt-6-luna-core-gpt-6-luna-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3","failure_count":0,"failure_rate":0.0,"input_tokens_total":536778,"known_provider_cost_quality_pareto":true,"latency_mean_ms":262.0940212340925,"latency_p50_ms":256.2655620058649,"latency_p95_ms":318.280074550421,"model_key":"jev-1.13.0","model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","normalized_quality_score":97.40259740259741,"observed_case_count":308,"observed_quality_latency_pareto":true,"output_tokens_total":256742,"primary_metric":"oos_detection_rate","primary_value":0.974025974025974,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.022544676000000003,"provider_cost_per_1k_cases_usd":0.07319700000000001,"provider_cost_per_case_usd":7.319700000000001e-05,"provider_cost_priced_cases":308,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"jev","processing":"standard","source":"TypeSafe AI public pricing: $42 per billion input tokens; output free.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":308,"provider_cost_total_usd":0.022544676000000003,"run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:consolidated:9468d75aebe9e805cefc3643f61a9e5fe6127c7a4e3a987608b30841b9bc9fa0","capability_id":"oos-calibration","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","failure_count":9,"failure_rate":0.05844155844155844,"input_tokens_total":102172,"known_provider_cost_quality_pareto":false,"latency_mean_ms":18534.834868738777,"latency_p50_ms":10469.861020508688,"latency_p95_ms":73937.57543984102,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","normalized_quality_score":95.45454545454545,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":113346,"primary_metric":"oos_detection_rate","primary_value":0.9545454545454546,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"lfm25-consolidated-latest","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":106024,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2420.62463879885,"latency_p50_ms":2395.2548329980345,"latency_p95_ms":2632.3175227080355,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","normalized_quality_score":96.1038961038961,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":3217,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":106024,"known_provider_cost_quality_pareto":false,"latency_mean_ms":3123.8922102271176,"latency_p50_ms":2692.1658124992973,"latency_p95_ms":4934.0581772929,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","normalized_quality_score":96.1038961038961,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":3335,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":93984,"known_provider_cost_quality_pareto":false,"latency_mean_ms":442.98370453869614,"latency_p50_ms":433.2297294968157,"latency_p95_ms":463.41581654851325,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","normalized_quality_score":50.0,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":2864,"primary_metric":"oos_detection_rate","primary_value":0.5,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","failure_count":0,"failure_rate":0.0,"input_tokens_total":93984,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1091.3614762078385,"latency_p50_ms":1085.873708507279,"latency_p95_ms":1192.1733147173654,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","normalized_quality_score":50.0,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":2894,"primary_metric":"oos_detection_rate","primary_value":0.5,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":93984,"known_provider_cost_quality_pareto":false,"latency_mean_ms":817.716475922101,"latency_p50_ms":811.0212289975607,"latency_p95_ms":881.2015722527576,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","normalized_quality_score":50.0,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":2960,"primary_metric":"oos_detection_rate","primary_value":0.5,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":93984,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1858.605324116637,"latency_p50_ms":1845.0444789996254,"latency_p95_ms":1977.098264250526,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","normalized_quality_score":97.40259740259741,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":2637,"primary_metric":"oos_detection_rate","primary_value":0.974025974025974,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","failure_count":0,"failure_rate":0.0,"input_tokens_total":93984,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2090.62336576844,"latency_p50_ms":2040.7832085038535,"latency_p95_ms":2489.353190278052,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","normalized_quality_score":98.05194805194806,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":2663,"primary_metric":"oos_detection_rate","primary_value":0.9805194805194806,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":93984,"known_provider_cost_quality_pareto":false,"latency_mean_ms":5469.462600662698,"latency_p50_ms":5220.9697289945325,"latency_p95_ms":8752.915718645818,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","normalized_quality_score":97.40259740259741,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":2600,"primary_metric":"oos_detection_rate","primary_value":0.974025974025974,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","failure_count":0,"failure_rate":0.0,"input_tokens_total":93984,"known_provider_cost_quality_pareto":false,"latency_mean_ms":3174.727481104624,"latency_p50_ms":3154.345499962801,"latency_p95_ms":3386.294052284211,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","normalized_quality_score":96.1038961038961,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":2629,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","failure_count":16,"failure_rate":0.1038961038961039,"input_tokens_total":84236,"known_provider_cost_quality_pareto":false,"latency_mean_ms":12654.062560350236,"latency_p50_ms":9886.834312499559,"latency_p95_ms":36285.84456239332,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","normalized_quality_score":81.81818181818183,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":2434,"primary_metric":"oos_detection_rate","primary_value":0.8181818181818182,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":108643,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2389.1731853761853,"latency_p50_ms":2335.455271000683,"latency_p95_ms":2798.3671812500684,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","normalized_quality_score":96.1038961038961,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":2950,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","sample_count":154}],"dataset_summaries":[{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":46999,"known_provider_cost_quality_pareto":false,"latency_mean_ms":8778.879600687424,"latency_p50_ms":8735.466374986572,"latency_p95_ms":9121.223125007236,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","normalized_quality_score":98.7012987012987,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1924,"primary_metric":"oos_correct","primary_value":0.987012987012987,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":46985,"known_provider_cost_quality_pareto":false,"latency_mean_ms":8656.081173180975,"latency_p50_ms":8601.950458993088,"latency_p95_ms":9165.10655801103,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","normalized_quality_score":90.9090909090909,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1710,"primary_metric":"oos_correct","primary_value":0.9090909090909091,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":77,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":136.1645048986153,"latency_p50_ms":126.8837500247173,"latency_p95_ms":177.19717521686113,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","normalized_quality_score":0.0,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":null,"primary_metric":"oos_correct","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":77,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":138.29759091921997,"latency_p50_ms":123.73316701268777,"latency_p95_ms":178.8445335580036,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","normalized_quality_score":0.0,"observed_case_count":77,"observed_quality_latency_pareto":true,"output_tokens_total":null,"primary_metric":"oos_correct","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":77,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":130.14289070005444,"latency_p50_ms":126.64366699755192,"latency_p95_ms":157.0738750160672,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","normalized_quality_score":0.0,"observed_case_count":77,"observed_quality_latency_pareto":true,"output_tokens_total":null,"primary_metric":"oos_correct","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.8-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":77,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":131.58038847810624,"latency_p50_ms":129.40495798829943,"latency_p95_ms":156.79174198303375,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","normalized_quality_score":0.0,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":null,"primary_metric":"oos_correct","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.8-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":4,"failure_rate":0.05194805194805195,"input_tokens_total":76515,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1893.9702813515466,"latency_p50_ms":1548.235541995382,"latency_p95_ms":3869.6360582049238,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","normalized_quality_score":94.8051948051948,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":3750,"primary_metric":"oos_correct","primary_value":0.948051948051948,"provider_cost_coverage_rate":0.948051948051948,"provider_cost_known":true,"provider_cost_observed_total_usd":0.019803,"provider_cost_per_1k_cases_usd":0.2712739726027398,"provider_cost_per_case_usd":0.00027127397260273976,"provider_cost_priced_cases":73,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":80699,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1190.0072024281699,"latency_p50_ms":1147.1069999970496,"latency_p95_ms":1587.2184588079108,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","normalized_quality_score":100.0,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1684,"primary_metric":"oos_correct","primary_value":1.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.018160600000000002,"provider_cost_per_1k_cases_usd":0.2358519480519481,"provider_cost_per_case_usd":0.0002358519480519481,"provider_cost_priced_cases":77,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":77,"provider_cost_total_usd":0.018160600000000002,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":1,"failure_rate":0.012987012987012988,"input_tokens_total":79656,"known_provider_cost_quality_pareto":false,"latency_mean_ms":3057.293945286187,"latency_p50_ms":2784.830249962397,"latency_p95_ms":4620.6940750125805,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","normalized_quality_score":97.40259740259741,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":5863,"primary_metric":"oos_correct","primary_value":0.974025974025974,"provider_cost_coverage_rate":0.987012987012987,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0229668,"provider_cost_per_1k_cases_usd":0.3021947368421053,"provider_cost_per_case_usd":0.0003021947368421053,"provider_cost_priced_cases":76,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":80699,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2182.9418576361804,"latency_p50_ms":1999.0120829897933,"latency_p95_ms":3477.565124980174,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","normalized_quality_score":100.0,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":2993,"primary_metric":"oos_correct","primary_value":1.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0197314,"provider_cost_per_1k_cases_usd":0.25625194805194806,"provider_cost_per_case_usd":0.0002562519480519481,"provider_cost_priced_cases":77,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":77,"provider_cost_total_usd":0.0197314,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":268460,"known_provider_cost_quality_pareto":true,"latency_mean_ms":261.2221466434829,"latency_p50_ms":256.74289550079266,"latency_p95_ms":320.20307664934074,"model_key":"jev-1.13.0","model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","normalized_quality_score":94.8051948051948,"observed_case_count":154,"observed_quality_latency_pareto":true,"output_tokens_total":128614,"primary_metric":"oos_correct","primary_value":0.948051948051948,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.01127532,"provider_cost_per_1k_cases_usd":0.07321636363636363,"provider_cost_per_case_usd":7.321636363636364e-05,"provider_cost_priced_cases":154,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"jev","processing":"standard","source":"TypeSafe AI public pricing: $42 per billion input tokens; output free.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":154,"provider_cost_total_usd":0.01127532,"sample_count":154},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":268318,"known_provider_cost_quality_pareto":true,"latency_mean_ms":262.96589582470205,"latency_p50_ms":256.0998745029792,"latency_p95_ms":316.27981425554026,"model_key":"jev-1.13.0","model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","normalized_quality_score":100.0,"observed_case_count":154,"observed_quality_latency_pareto":true,"output_tokens_total":128128,"primary_metric":"oos_correct","primary_value":1.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.011269356000000001,"provider_cost_per_1k_cases_usd":0.07317763636363636,"provider_cost_per_case_usd":7.317763636363637e-05,"provider_cost_priced_cases":154,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"jev","processing":"standard","source":"TypeSafe AI public pricing: $42 per billion input tokens; output free.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":154,"provider_cost_total_usd":0.011269356000000001,"sample_count":154},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":8,"failure_rate":0.1038961038961039,"input_tokens_total":48636,"known_provider_cost_quality_pareto":false,"latency_mean_ms":25299.56137446094,"latency_p50_ms":12379.735084017739,"latency_p95_ms":74805.01954995561,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","normalized_quality_score":85.71428571428571,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":70948,"primary_metric":"oos_correct","primary_value":0.8571428571428571,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":1,"failure_rate":0.012987012987012988,"input_tokens_total":53536,"known_provider_cost_quality_pareto":false,"latency_mean_ms":11770.108363016618,"latency_p50_ms":9689.080916927196,"latency_p95_ms":24273.661758471288,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","normalized_quality_score":94.8051948051948,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":42398,"primary_metric":"oos_correct","primary_value":0.948051948051948,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":53013,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2499.965928078216,"latency_p50_ms":2489.686166998581,"latency_p95_ms":2695.2199669991387,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","normalized_quality_score":98.7012987012987,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1799,"primary_metric":"oos_correct","primary_value":0.987012987012987,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":53011,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2341.2833495194836,"latency_p50_ms":2314.8674159892835,"latency_p95_ms":2477.2517746081576,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","normalized_quality_score":93.5064935064935,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1418,"primary_metric":"oos_correct","primary_value":0.935064935064935,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":53013,"known_provider_cost_quality_pareto":false,"latency_mean_ms":3766.431912324541,"latency_p50_ms":3453.931582989753,"latency_p95_ms":5078.716925196933,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","normalized_quality_score":98.7012987012987,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1832,"primary_metric":"oos_correct","primary_value":0.987012987012987,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":53011,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2481.3525081296943,"latency_p50_ms":2494.065542006865,"latency_p95_ms":2691.221599999699,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","normalized_quality_score":93.5064935064935,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1503,"primary_metric":"oos_correct","primary_value":0.935064935064935,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":46999,"known_provider_cost_quality_pareto":false,"latency_mean_ms":431.5476136492995,"latency_p50_ms":428.76354200416245,"latency_p95_ms":454.68750780564733,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","normalized_quality_score":100.0,"observed_case_count":77,"observed_quality_latency_pareto":true,"output_tokens_total":1448,"primary_metric":"oos_correct","primary_value":1.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":46985,"known_provider_cost_quality_pareto":false,"latency_mean_ms":454.4197954280928,"latency_p50_ms":437.74258300254587,"latency_p95_ms":470.6647918006638,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","normalized_quality_score":0.0,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1416,"primary_metric":"oos_correct","primary_value":0.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":46999,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1085.9827381821785,"latency_p50_ms":1085.135250003077,"latency_p95_ms":1179.1477915947328,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","normalized_quality_score":100.0,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1451,"primary_metric":"oos_correct","primary_value":1.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":46985,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1096.7402142334984,"latency_p50_ms":1086.6620419546962,"latency_p95_ms":1191.4266084204428,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","normalized_quality_score":0.0,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1443,"primary_metric":"oos_correct","primary_value":0.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":46999,"known_provider_cost_quality_pareto":false,"latency_mean_ms":831.4678630655156,"latency_p50_ms":826.020000007702,"latency_p95_ms":921.0994412045693,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","normalized_quality_score":100.0,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1503,"primary_metric":"oos_correct","primary_value":1.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":46985,"known_provider_cost_quality_pareto":false,"latency_mean_ms":803.9650887786864,"latency_p50_ms":799.419791001128,"latency_p95_ms":844.2399912077235,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","normalized_quality_score":0.0,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1457,"primary_metric":"oos_correct","primary_value":0.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":46999,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1897.1279539875065,"latency_p50_ms":1879.2896250088234,"latency_p95_ms":2049.794733597082,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","normalized_quality_score":98.7012987012987,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1447,"primary_metric":"oos_correct","primary_value":0.987012987012987,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":46985,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1820.082694245767,"latency_p50_ms":1807.692207992659,"latency_p95_ms":1940.8644089999143,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","normalized_quality_score":96.1038961038961,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1190,"primary_metric":"oos_correct","primary_value":0.961038961038961,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":46999,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2208.720919327635,"latency_p50_ms":2165.3660840238445,"latency_p95_ms":2628.3127831760794,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","normalized_quality_score":98.7012987012987,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1488,"primary_metric":"oos_correct","primary_value":0.987012987012987,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":46985,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1972.5258122092453,"latency_p50_ms":1946.7772909556516,"latency_p95_ms":2136.2961751641706,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","normalized_quality_score":97.40259740259741,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1175,"primary_metric":"oos_correct","primary_value":0.974025974025974,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":46999,"known_provider_cost_quality_pareto":false,"latency_mean_ms":4758.679551974259,"latency_p50_ms":3863.904249999905,"latency_p95_ms":7737.850358401193,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","normalized_quality_score":97.40259740259741,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1436,"primary_metric":"oos_correct","primary_value":0.974025974025974,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":46985,"known_provider_cost_quality_pareto":false,"latency_mean_ms":6180.245649351136,"latency_p50_ms":6215.419833009946,"latency_p95_ms":8815.064166203956,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","normalized_quality_score":97.40259740259741,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1164,"primary_metric":"oos_correct","primary_value":0.974025974025974,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":46999,"known_provider_cost_quality_pareto":false,"latency_mean_ms":3249.9961693923883,"latency_p50_ms":3231.6983329947107,"latency_p95_ms":3474.731591984164,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","normalized_quality_score":98.7012987012987,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1435,"primary_metric":"oos_correct","primary_value":0.987012987012987,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":46985,"known_provider_cost_quality_pareto":false,"latency_mean_ms":3099.45879281686,"latency_p50_ms":3090.7840419677086,"latency_p95_ms":3210.958441812545,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","normalized_quality_score":93.5064935064935,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1194,"primary_metric":"oos_correct","primary_value":0.935064935064935,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":3,"failure_rate":0.03896103896103896,"input_tokens_total":45176,"known_provider_cost_quality_pareto":false,"latency_mean_ms":11085.036545479934,"latency_p50_ms":10003.22662499093,"latency_p95_ms":10811.810099994182,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","normalized_quality_score":93.5064935064935,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1399,"primary_metric":"oos_correct","primary_value":0.935064935064935,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":13,"failure_rate":0.16883116883116883,"input_tokens_total":39060,"known_provider_cost_quality_pareto":false,"latency_mean_ms":14223.088575220538,"latency_p50_ms":9674.696875008522,"latency_p95_ms":36420.40486720216,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","normalized_quality_score":66.23376623376623,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1035,"primary_metric":"oos_correct","primary_value":0.6623376623376623,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":54325,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2492.3314697267156,"latency_p50_ms":2434.8514579905896,"latency_p95_ms":2920.9799582022247,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","normalized_quality_score":96.1038961038961,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1631,"primary_metric":"oos_correct","primary_value":0.961038961038961,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":54318,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2286.0149010256546,"latency_p50_ms":2229.7938340052497,"latency_p95_ms":2551.516925197211,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","normalized_quality_score":96.1038961038961,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1319,"primary_metric":"oos_correct","primary_value":0.961038961038961,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77}]},"disagreements":[],"family_breakdown":[],"schema_version":"2"},"qa-abstention":{"benchmark_signatures":["sha256:benchmark:1fa41ef05579230896c951f4e01da57ccf719647ad6f8235008243556409a7cb","sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","sha256:consolidated:b832fe612fcfe138845fe193a557c686f68ecb2f59d1b3f4784e653e9ec82e5c"],"capability_id":"qa-abstention","cases":[],"cells":[{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-06T09:38:49.212082+00:00","deployment":"local","effective_model_id":"bonsai-27b-q10","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"profile":"core","provider_key":"korgis","quantization":"Q1_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"bonsai-27b-q10","process_cpu_percent_avg":9.839476147142088,"process_rss_bytes_avg":6890131131.833357,"process_rss_bytes_peak":10222600192.0,"run_id":"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core","sample_count":690,"sample_interval_ms":311.07264410993747,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5791531008.0},"run_id":"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T13:59:29.849450+00:00","deployment":"api","effective_model_id":"gemini-3.5-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.5","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.016666666666666666,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core","runtime_key":"gemini-api","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T13:38:52.091017+00:00","deployment":"api","effective_model_id":"gemini-3.8-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.8","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.8-flash","model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.0,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core","runtime_key":"gemini-api","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-06T12:16:26.396465+00:00","deployment":"api","effective_model_id":"gpt-5.6-luna","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core","runtime_key":"openai-api","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T13:12:07.696210+00:00","deployment":"api","effective_model_id":"gpt-6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","experiment_kind":"standard","failure_count":0,"family":"gpt-6","generation_policy":{},"git_commit":"7f796c10d99e934da6417ea54405fbf9352ad399","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.95,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"gpt-6-luna-core-gpt-6-luna-qa-abstention-core","runtime_key":"openai-api","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":10947100928,"benchmark_signature":"sha256:consolidated:b832fe612fcfe138845fe193a557c686f68ecb2f59d1b3f4784e653e9ec82e5c","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-22T13:12:52.242438+00:00","deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","experiment_kind":"consolidated","failure_count":0,"family":"lfm2.5","generation_policy":{"generation_configs":[{"extra":{},"max_output_tokens":2048,"seed":42,"stop":[],"temperature":0.0},{"extra":{},"max_output_tokens":4096,"seed":42,"stop":[],"temperature":0.0}],"mixed_generation":true},"git_commit":null,"model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"profile":"core","provider_key":"korgis","quantization":"UD-Q3_K_XL","resource_summary":null,"run_id":"lfm25-consolidated-latest","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:1fa41ef05579230896c951f4e01da57ccf719647ad6f8235008243556409a7cb","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-20T10:03:44.604674+00:00","deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","experiment_kind":"standard","failure_count":0,"family":"limite","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","parameters_b":1.0,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.0,"profile":"smoke","provider_key":"transformers","quantization":null,"resource_summary":null,"run_id":"b6b4123a-1d20-4c39-8644-ea00c1e5893b","runtime_key":"transformers-local","sample_count":12,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T11:57:09.469183+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b","process_cpu_percent_avg":9.533096212147692,"process_rss_bytes_avg":7103316869.12,"process_rss_bytes_peak":11089903616.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","sample_count":264,"sample_interval_ms":273.58964924757555,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6888439808.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T13:14:43.428857+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b-q8","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b-q8","process_cpu_percent_avg":7.166483605707706,"process_rss_bytes_avg":6463349882.22984,"process_rss_bytes_peak":10168467456.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core","sample_count":359,"sample_interval_ms":297.72169937988014,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5578047488.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T13:35:56.623493+00:00","deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.48333333333333334,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-0.8b-q4km","process_cpu_percent_avg":16.023245275126758,"process_rss_bytes_avg":2421507094.7555556,"process_rss_bytes_peak":3926736896.0,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core","sample_count":126,"sample_interval_ms":259.9318482166079,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":14612348928.0},"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":3775709216,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T11:22:09.705119+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-bf16","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","parameters_b":2.0,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.6333333333333333,"profile":"core","provider_key":"korgis","quantization":"BF16","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","model_key":"qwen3.5-2b-bf16","process_cpu_percent_avg":5.5345916105632815,"process_rss_bytes_avg":5136314017.564444,"process_rss_bytes_peak":5894602752.0,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core","sample_count":226,"sample_interval_ms":279.53779863981055,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":7390167040.0},"run_id":"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T11:22:51.387753+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.7833333333333333,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-2b-q4km","process_cpu_percent_avg":14.926939453226597,"process_rss_bytes_avg":2006719510.7555552,"process_rss_bytes_peak":2334687232.0,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core","sample_count":182,"sample_interval_ms":249.7748705191043,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6527320064.0},"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T16:36:39.120802+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"cb061291446b8dbd79f524a4fefcd4a4406e2a0d","model_id":"unsloth/qwen3.5-4b","model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-4b-q4km","process_cpu_percent_avg":9.78397589283007,"process_rss_bytes_avg":7517191058.123175,"process_rss_bytes_peak":10744905728.0,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":261,"sample_interval_ms":276.8212083626713,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6344114176.0},"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":4482403488,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T11:31:50.661853+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q8","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","parameters_b":4.0,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","model_key":"qwen3.5-4b-q8","process_cpu_percent_avg":6.023120834777095,"process_rss_bytes_avg":7696510827.330371,"process_rss_bytes_peak":10978181120.0,"run_id":"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core","sample_count":278,"sample_interval_ms":297.275809231578,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":7347224576.0},"run_id":"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T12:38:16.531029+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.9333333333333333,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-9b-q4km","process_cpu_percent_avg":6.3201481230234045,"process_rss_bytes_avg":8711409521.940317,"process_rss_bytes_peak":11538563072.0,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","sample_count":345,"sample_interval_ms":281.76636477524414,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":7018201088.0},"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":4121781472,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T11:47:21.635679+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-ud-q2kxl","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"qwen/qwen3.5-9b-ud-q2kxl","model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","parameters_b":9.0,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"profile":"core","provider_key":"korgis","quantization":"UD-Q2_K_XL","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","model_key":"qwen3.5-9b-ud-q2kxl","process_cpu_percent_avg":4.725490643337322,"process_rss_bytes_avg":7851091423.1669855,"process_rss_bytes_peak":11102388224.0,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core","sample_count":336,"sample_interval_ms":289.6220233097523,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":8726102016.0},"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":6192222208,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-06T14:37:09.228523+00:00","deployment":"local","effective_model_id":"qwen3.8-27b-iq1s","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","experiment_kind":"standard","failure_count":0,"family":"qwen3.8","generation_policy":{},"git_commit":"640138c3429882a42797d28074f98d3fea98a63b","model_id":"Qwen/Qwen3.8-27B","model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","parameters_b":27.0,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8,"profile":"core","provider_key":"korgis","quantization":"UD-IQ1_S","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","model_key":"qwen3.8-27b-iq1s","process_cpu_percent_avg":5.983418081521953,"process_rss_bytes_avg":9550237266.862207,"process_rss_bytes_peak":13159825408.0,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","sample_count":792,"sample_interval_ms":301.02958077118194,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5182652416.0},"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-06T09:01:09.715786+00:00","deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"xhtoken/spark-x2.5-4b","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.95,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"spark-x2.5-4b-q4km","process_cpu_percent_avg":9.07249993238987,"process_rss_bytes_avg":3570442922.666667,"process_rss_bytes_peak":4295933952.0,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core","sample_count":244,"sample_interval_ms":280.7728714734669,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6034112512.0},"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"}],"decision":{"capability_summaries":[{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":3175.179379200805,"latency_p50_ms":3054.529208500753,"latency_p95_ms":3838.2669604550756,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","normalized_quality_score":86.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1305,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":59,"failure_rate":0.9833333333333333,"input_tokens_total":120,"known_provider_cost_quality_pareto":false,"latency_mean_ms":244.35933958253977,"latency_p50_ms":190.14506248640828,"latency_p95_ms":281.4437542168883,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","normalized_quality_score":1.6666666666666667,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":15,"primary_metric":"qa_correct","primary_value":0.016666666666666666,"provider_cost_coverage_rate":0.016666666666666666,"provider_cost_known":true,"provider_cost_observed_total_usd":1.35e-05,"provider_cost_per_1k_cases_usd":0.0135,"provider_cost_per_case_usd":1.35e-05,"provider_cost_priced_cases":1,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":60,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":136.61597294752332,"latency_p50_ms":129.0290629840456,"latency_p95_ms":191.78036249941215,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","normalized_quality_score":0.0,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":null,"primary_metric":"qa_correct","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.8-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","failure_count":0,"failure_rate":0.0,"input_tokens_total":9763,"known_provider_cost_quality_pareto":true,"latency_mean_ms":1337.4706270170766,"latency_p50_ms":1291.5158120013075,"latency_p95_ms":1930.584752104187,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","normalized_quality_score":86.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1370,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0035966,"provider_cost_per_1k_cases_usd":0.059943333333333335,"provider_cost_per_case_usd":5.994333333333333e-05,"provider_cost_priced_cases":60,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":60,"provider_cost_total_usd":0.0035966,"run_id":"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":0,"failure_rate":0.0,"input_tokens_total":9763,"known_provider_cost_quality_pareto":true,"latency_mean_ms":1647.2096979327034,"latency_p50_ms":1527.1968334855046,"latency_p95_ms":2859.7770288557504,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","normalized_quality_score":95.0,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1869,"primary_metric":"qa_correct","primary_value":0.95,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0041954,"provider_cost_per_1k_cases_usd":0.06992333333333334,"provider_cost_per_case_usd":6.992333333333334e-05,"provider_cost_priced_cases":60,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":60,"provider_cost_total_usd":0.0041954,"run_id":"gpt-6-luna-core-gpt-6-luna-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:consolidated:b832fe612fcfe138845fe193a557c686f68ecb2f59d1b3f4784e653e9ec82e5c","capability_id":"qa-abstention","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","failure_count":0,"failure_rate":0.0,"input_tokens_total":8170,"known_provider_cost_quality_pareto":false,"latency_mean_ms":8713.557179173222,"latency_p50_ms":7030.449937446974,"latency_p95_ms":18365.44755841605,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","normalized_quality_score":96.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":23009,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"lfm25-consolidated-latest","sample_count":60},{"benchmark_signature":"sha256:benchmark:1fa41ef05579230896c951f4e01da57ccf719647ad6f8235008243556409a7cb","capability_id":"qa-abstention","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","failure_count":12,"failure_rate":1.0,"input_tokens_total":803,"known_provider_cost_quality_pareto":false,"latency_mean_ms":44512.825486056194,"latency_p50_ms":50686.88316695625,"latency_p95_ms":53967.651789076626,"model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","normalized_quality_score":0.0,"observed_case_count":12,"observed_quality_latency_pareto":false,"output_tokens_total":20683,"primary_metric":"qa_correct","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":12,"provider_cost_total_usd":null,"run_id":"b6b4123a-1d20-4c39-8644-ea00c1e5893b","sample_count":12},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":8421,"known_provider_cost_quality_pareto":false,"latency_mean_ms":860.4363417003091,"latency_p50_ms":849.5050004930818,"latency_p95_ms":1040.7841770924276,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","normalized_quality_score":96.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":897,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":8421,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1412.304399300774,"latency_p50_ms":1362.8550000066753,"latency_p95_ms":1983.9075827017948,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","normalized_quality_score":96.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":877,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":212.94066396609801,"latency_p50_ms":204.67102049587993,"latency_p95_ms":269.198137501371,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","normalized_quality_score":48.333333333333336,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":999,"primary_metric":"qa_correct","primary_value":0.48333333333333334,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","failure_count":10,"failure_rate":0.16666666666666666,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":665.2421306236647,"latency_p50_ms":616.8598124640994,"latency_p95_ms":1006.9639482157072,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","normalized_quality_score":63.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1014,"primary_metric":"qa_correct","primary_value":0.6333333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":423.106404900318,"latency_p50_ms":398.915562494949,"latency_p95_ms":549.0888670014099,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","normalized_quality_score":78.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":1025,"primary_metric":"qa_correct","primary_value":0.7833333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":876.0588624500087,"latency_p50_ms":786.4938955026446,"latency_p95_ms":1081.4400829891365,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","normalized_quality_score":86.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1085,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":989.199661828267,"latency_p50_ms":908.3277915196959,"latency_p95_ms":1313.4596493822753,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","normalized_quality_score":86.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1036,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1270.930086817437,"latency_p50_ms":1220.2731874931487,"latency_p95_ms":1647.5607910033425,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","normalized_quality_score":93.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":930,"primary_metric":"qa_correct","primary_value":0.9333333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1255.4060783363336,"latency_p50_ms":1200.648166501196,"latency_p95_ms":1755.204953884822,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","normalized_quality_score":86.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1014,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","failure_count":1,"failure_rate":0.016666666666666666,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":3615.463377833172,"latency_p50_ms":3467.9861039985553,"latency_p95_ms":4320.6283913576035,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","normalized_quality_score":80.0,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":961,"primary_metric":"qa_correct","primary_value":0.8,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":8552,"known_provider_cost_quality_pareto":false,"latency_mean_ms":772.222803482388,"latency_p50_ms":746.6509164951276,"latency_p95_ms":985.1788809428398,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","normalized_quality_score":95.0,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":871,"primary_metric":"qa_correct","primary_value":0.95,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core","sample_count":60}],"dataset_summaries":[{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":3175.179379200805,"latency_p50_ms":3054.529208500753,"latency_p95_ms":3838.2669604550756,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","normalized_quality_score":86.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1305,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":59,"failure_rate":0.9833333333333333,"input_tokens_total":120,"known_provider_cost_quality_pareto":false,"latency_mean_ms":244.35933958253977,"latency_p50_ms":190.14506248640828,"latency_p95_ms":281.4437542168883,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","normalized_quality_score":1.6666666666666667,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":15,"primary_metric":"qa_correct","primary_value":0.016666666666666666,"provider_cost_coverage_rate":0.016666666666666666,"provider_cost_known":true,"provider_cost_observed_total_usd":1.35e-05,"provider_cost_per_1k_cases_usd":0.0135,"provider_cost_per_case_usd":1.35e-05,"provider_cost_priced_cases":1,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":60,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":136.61597294752332,"latency_p50_ms":129.0290629840456,"latency_p95_ms":191.78036249941215,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","normalized_quality_score":0.0,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":null,"primary_metric":"qa_correct","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.8-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":9763,"known_provider_cost_quality_pareto":true,"latency_mean_ms":1337.4706270170766,"latency_p50_ms":1291.5158120013075,"latency_p95_ms":1930.584752104187,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","normalized_quality_score":86.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1370,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0035966,"provider_cost_per_1k_cases_usd":0.059943333333333335,"provider_cost_per_case_usd":5.994333333333333e-05,"provider_cost_priced_cases":60,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":60,"provider_cost_total_usd":0.0035966,"sample_count":60},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":9763,"known_provider_cost_quality_pareto":true,"latency_mean_ms":1647.2096979327034,"latency_p50_ms":1527.1968334855046,"latency_p95_ms":2859.7770288557504,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","normalized_quality_score":95.0,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1869,"primary_metric":"qa_correct","primary_value":0.95,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0041954,"provider_cost_per_1k_cases_usd":0.06992333333333334,"provider_cost_per_case_usd":6.992333333333334e-05,"provider_cost_priced_cases":60,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":60,"provider_cost_total_usd":0.0041954,"sample_count":60},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":8170,"known_provider_cost_quality_pareto":false,"latency_mean_ms":8713.557179173222,"latency_p50_ms":7030.449937446974,"latency_p95_ms":18365.44755841605,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","normalized_quality_score":96.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":23009,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":12,"failure_rate":1.0,"input_tokens_total":803,"known_provider_cost_quality_pareto":false,"latency_mean_ms":44512.825486056194,"latency_p50_ms":50686.88316695625,"latency_p95_ms":53967.651789076626,"model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","normalized_quality_score":0.0,"observed_case_count":12,"observed_quality_latency_pareto":false,"output_tokens_total":20683,"primary_metric":"qa_correct","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":12,"provider_cost_total_usd":null,"sample_count":12},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":8421,"known_provider_cost_quality_pareto":false,"latency_mean_ms":860.4363417003091,"latency_p50_ms":849.5050004930818,"latency_p95_ms":1040.7841770924276,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","normalized_quality_score":96.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":897,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":8421,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1412.304399300774,"latency_p50_ms":1362.8550000066753,"latency_p95_ms":1983.9075827017948,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","normalized_quality_score":96.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":877,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":212.94066396609801,"latency_p50_ms":204.67102049587993,"latency_p95_ms":269.198137501371,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","normalized_quality_score":48.333333333333336,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":999,"primary_metric":"qa_correct","primary_value":0.48333333333333334,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":10,"failure_rate":0.16666666666666666,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":665.2421306236647,"latency_p50_ms":616.8598124640994,"latency_p95_ms":1006.9639482157072,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","normalized_quality_score":63.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1014,"primary_metric":"qa_correct","primary_value":0.6333333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":423.106404900318,"latency_p50_ms":398.915562494949,"latency_p95_ms":549.0888670014099,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","normalized_quality_score":78.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":1025,"primary_metric":"qa_correct","primary_value":0.7833333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":876.0588624500087,"latency_p50_ms":786.4938955026446,"latency_p95_ms":1081.4400829891365,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","normalized_quality_score":86.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1085,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":989.199661828267,"latency_p50_ms":908.3277915196959,"latency_p95_ms":1313.4596493822753,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","normalized_quality_score":86.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1036,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1270.930086817437,"latency_p50_ms":1220.2731874931487,"latency_p95_ms":1647.5607910033425,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","normalized_quality_score":93.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":930,"primary_metric":"qa_correct","primary_value":0.9333333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1255.4060783363336,"latency_p50_ms":1200.648166501196,"latency_p95_ms":1755.204953884822,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","normalized_quality_score":86.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1014,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":1,"failure_rate":0.016666666666666666,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":3615.463377833172,"latency_p50_ms":3467.9861039985553,"latency_p95_ms":4320.6283913576035,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","normalized_quality_score":80.0,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":961,"primary_metric":"qa_correct","primary_value":0.8,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":8552,"known_provider_cost_quality_pareto":false,"latency_mean_ms":772.222803482388,"latency_p50_ms":746.6509164951276,"latency_p95_ms":985.1788809428398,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","normalized_quality_score":95.0,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":871,"primary_metric":"qa_correct","primary_value":0.95,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60}]},"disagreements":[],"family_breakdown":[],"schema_version":"2"},"structured-output":{"benchmark_signatures":["sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","sha256:consolidated:fd230e4e9f030196ab13d72a0beb260f92fbe3bfb1cfa331ed8e0678b55123f9"],"capability_id":"structured-output","cases":[],"cells":[{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-06T09:35:22.771531+00:00","deployment":"local","effective_model_id":"bonsai-27b-q10","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7833333333333333,"profile":"core","provider_key":"korgis","quantization":"Q1_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"bonsai-27b-q10","process_cpu_percent_avg":9.52518999663954,"process_rss_bytes_avg":6054163284.918202,"process_rss_bytes_peak":8022573056.0,"run_id":"2026-10-core-v1-bonsai-27b-q10-structured-output-core","sample_count":990,"sample_interval_ms":317.9005326878062,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5717721088.0},"run_id":"2026-10-core-v1-bonsai-27b-q10-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T13:59:13.872786+00:00","deployment":"api","effective_model_id":"gemini-3.5-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.5","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.03333333333333333,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core","runtime_key":"gemini-api","sample_count":60,"task_id":"structured-output"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T13:38:42.203597+00:00","deployment":"api","effective_model_id":"gemini-3.8-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.8","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.8-flash","model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.0,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core","runtime_key":"gemini-api","sample_count":60,"task_id":"structured-output"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-06T12:15:04.761551+00:00","deployment":"api","effective_model_id":"gpt-5.6-luna","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.9166666666666666,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-structured-output-core","runtime_key":"openai-api","sample_count":60,"task_id":"structured-output"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T13:10:27.524208+00:00","deployment":"api","effective_model_id":"gpt-6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","experiment_kind":"standard","failure_count":0,"family":"gpt-6","generation_policy":{},"git_commit":"7f796c10d99e934da6417ea54405fbf9352ad399","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.8333333333333334,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"gpt-6-luna-core-gpt-6-luna-structured-output-core","runtime_key":"openai-api","sample_count":60,"task_id":"structured-output"},{"artifact_format":"gguf","artifact_size_bytes":10947100928,"benchmark_signature":"sha256:consolidated:fd230e4e9f030196ab13d72a0beb260f92fbe3bfb1cfa331ed8e0678b55123f9","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-22T13:12:52.242438+00:00","deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","experiment_kind":"consolidated","failure_count":0,"family":"lfm2.5","generation_policy":{"generation_configs":[{"extra":{},"max_output_tokens":2048,"seed":42,"stop":[],"temperature":0.0},{"extra":{},"max_output_tokens":4096,"seed":42,"stop":[],"temperature":0.0}],"mixed_generation":true},"git_commit":null,"model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7666666666666667,"profile":"core","provider_key":"korgis","quantization":"UD-Q3_K_XL","resource_summary":null,"run_id":"lfm25-consolidated-latest","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T11:56:03.800158+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7333333333333333,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b","process_cpu_percent_avg":8.190733436222855,"process_rss_bytes_avg":4822041698.065609,"process_rss_bytes_peak":6531989504.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-structured-output-core","sample_count":423,"sample_interval_ms":283.9724207012575,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5921701888.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T13:13:01.309902+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b-q8","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7166666666666667,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b-q8","process_cpu_percent_avg":4.8684869265121,"process_rss_bytes_avg":5683850551.952017,"process_rss_bytes_peak":6735822848.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core","sample_count":654,"sample_interval_ms":307.429391476698,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5616762880.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T13:35:30.102922+00:00","deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.6833333333333333,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-0.8b-q4km","process_cpu_percent_avg":20.24050747900639,"process_rss_bytes_avg":2703435980.8,"process_rss_bytes_peak":4575264768.0,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core","sample_count":195,"sample_interval_ms":237.40003127216673,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":13519536128.0},"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"},{"artifact_format":"gguf","artifact_size_bytes":3775709216,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T11:21:16.650276+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-bf16","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","parameters_b":2.0,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7166666666666667,"profile":"core","provider_key":"korgis","quantization":"BF16","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","model_key":"qwen3.5-2b-bf16","process_cpu_percent_avg":5.386938078405328,"process_rss_bytes_avg":4179186975.941509,"process_rss_bytes_peak":4437606400.0,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core","sample_count":391,"sample_interval_ms":306.911332594406,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4923883520.0},"run_id":"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T10:55:49.921242+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7333333333333333,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-2b-q4km","process_cpu_percent_avg":15.462938913292911,"process_rss_bytes_avg":2241664864.711111,"process_rss_bytes_peak":2811346944.0,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core","sample_count":253,"sample_interval_ms":275.0343119529134,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6283788288.0},"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T16:36:39.120802+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"cb061291446b8dbd79f524a4fefcd4a4406e2a0d","model_id":"unsloth/qwen3.5-4b","model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7333333333333333,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-4b-q4km","process_cpu_percent_avg":8.486342485499302,"process_rss_bytes_avg":5121284030.324117,"process_rss_bytes_peak":6889406464.0,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":411,"sample_interval_ms":286.355363495796,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5873369088.0},"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"},{"artifact_format":"gguf","artifact_size_bytes":4482403488,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T11:30:38.611895+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q8","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","parameters_b":4.0,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7333333333333333,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","model_key":"qwen3.5-4b-q8","process_cpu_percent_avg":5.698576220486698,"process_rss_bytes_avg":6661104577.102827,"process_rss_bytes_peak":9700769792.0,"run_id":"qwen-variants-core-qwen3.5-4b-q8-structured-output-core","sample_count":477,"sample_interval_ms":317.01940462127334,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5716574208.0},"run_id":"qwen-variants-core-qwen3.5-4b-q8-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T12:36:44.930375+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7666666666666667,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-9b-q4km","process_cpu_percent_avg":5.8239325570441824,"process_rss_bytes_avg":6475378858.969259,"process_rss_bytes_peak":7243612160.0,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core","sample_count":617,"sample_interval_ms":298.72389312365516,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5452595200.0},"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"},{"artifact_format":"gguf","artifact_size_bytes":4121781472,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T11:45:53.885678+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-ud-q2kxl","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"qwen/qwen3.5-9b-ud-q2kxl","model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","parameters_b":9.0,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7166666666666667,"profile":"core","provider_key":"korgis","quantization":"UD-Q2_K_XL","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","model_key":"qwen3.5-9b-ud-q2kxl","process_cpu_percent_avg":4.974081455251981,"process_rss_bytes_avg":7070526205.54296,"process_rss_bytes_peak":10778755072.0,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core","sample_count":559,"sample_interval_ms":311.91939930876663,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5600329728.0},"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"},{"artifact_format":"gguf","artifact_size_bytes":6192222208,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-06T14:37:06.404488+00:00","deployment":"local","effective_model_id":"qwen3.8-27b-iq1s","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","experiment_kind":"standard","failure_count":0,"family":"qwen3.8","generation_policy":{},"git_commit":"640138c3429882a42797d28074f98d3fea98a63b","model_id":"Qwen/Qwen3.8-27B","model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","parameters_b":27.0,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.6833333333333333,"profile":"core","provider_key":"korgis","quantization":"UD-IQ1_S","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","model_key":"qwen3.8-27b-iq1s","process_cpu_percent_avg":5.429397449534366,"process_rss_bytes_avg":7678658248.931642,"process_rss_bytes_peak":9314566144.0,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","sample_count":1314,"sample_interval_ms":323.2460322738765,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4307910656.0},"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-06T09:00:08.668486+00:00","deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"xhtoken/spark-x2.5-4b","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7666666666666667,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"spark-x2.5-4b-q4km","process_cpu_percent_avg":7.982890411325502,"process_rss_bytes_avg":3362284766.8271284,"process_rss_bytes_peak":3593027584.0,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core","sample_count":417,"sample_interval_ms":298.61884293800324,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5815320576.0},"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"}],"decision":{"capability_summaries":[{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":6,"failure_rate":0.1,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":4839.719958333929,"latency_p50_ms":4484.091166501457,"latency_p95_ms":7536.193280648876,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","normalized_quality_score":78.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2689,"primary_metric":"exact_match","primary_value":0.7833333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-bonsai-27b-q10-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":58,"failure_rate":0.9666666666666667,"input_tokens_total":242,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1515.5912111426005,"latency_p50_ms":160.70060449419543,"latency_p95_ms":2834.6031663066206,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","normalized_quality_score":3.3333333333333335,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":64,"primary_metric":"exact_match","primary_value":0.03333333333333333,"provider_cost_coverage_rate":0.03333333333333333,"provider_cost_known":true,"provider_cost_observed_total_usd":3.735e-05,"provider_cost_per_1k_cases_usd":0.018675,"provider_cost_per_case_usd":1.8675e-05,"provider_cost_priced_cases":2,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":60,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":4986.970465994091,"latency_p50_ms":192.07681200350635,"latency_p95_ms":60002.98497467884,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","normalized_quality_score":0.0,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":null,"primary_metric":"exact_match","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.8-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","failure_count":0,"failure_rate":0.0,"input_tokens_total":10570,"known_provider_cost_quality_pareto":true,"latency_mean_ms":1934.9246188348236,"latency_p50_ms":1730.0426670044544,"latency_p95_ms":2594.3603166953817,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","normalized_quality_score":91.66666666666666,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":3316,"primary_metric":"exact_match","primary_value":0.9166666666666666,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0060932,"provider_cost_per_1k_cases_usd":0.10155333333333333,"provider_cost_per_case_usd":0.00010155333333333333,"provider_cost_priced_cases":60,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":60,"provider_cost_total_usd":0.0060932,"run_id":"2026-10-core-v1-gpt-5.6-luna-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":0,"failure_rate":0.0,"input_tokens_total":10570,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2258.3810763889537,"latency_p50_ms":2046.710938011529,"latency_p95_ms":3548.6162735644007,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","normalized_quality_score":83.33333333333334,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":3356,"primary_metric":"exact_match","primary_value":0.8333333333333334,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0061412,"provider_cost_per_1k_cases_usd":0.10235333333333334,"provider_cost_per_case_usd":0.00010235333333333333,"provider_cost_priced_cases":60,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":60,"provider_cost_total_usd":0.0061412,"run_id":"gpt-6-luna-core-gpt-6-luna-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:consolidated:fd230e4e9f030196ab13d72a0beb260f92fbe3bfb1cfa331ed8e0678b55123f9","capability_id":"structured-output","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","failure_count":0,"failure_rate":0.0,"input_tokens_total":8212,"known_provider_cost_quality_pareto":false,"latency_mean_ms":11260.75794856879,"latency_p50_ms":9417.586916533764,"latency_p95_ms":22641.866754152565,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","normalized_quality_score":76.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":35103,"primary_metric":"exact_match","primary_value":0.7666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"lfm25-consolidated-latest","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":2,"failure_rate":0.03333333333333333,"input_tokens_total":8424,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1666.822604850798,"latency_p50_ms":1567.1223544995883,"latency_p95_ms":2798.8641288560757,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","normalized_quality_score":73.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2345,"primary_metric":"exact_match","primary_value":0.7333333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":2,"failure_rate":0.03333333333333333,"input_tokens_total":8424,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2980.6488194495614,"latency_p50_ms":2663.501458497194,"latency_p95_ms":5424.1780169439135,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","normalized_quality_score":71.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2346,"primary_metric":"exact_match","primary_value":0.7166666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":3,"failure_rate":0.05,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":473.98608951771166,"latency_p50_ms":432.87633300496964,"latency_p95_ms":798.6456500082567,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","normalized_quality_score":68.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":2661,"primary_metric":"exact_match","primary_value":0.6833333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","failure_count":5,"failure_rate":0.08333333333333333,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1611.7719881663409,"latency_p50_ms":1479.928854474565,"latency_p95_ms":3047.5117707916065,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","normalized_quality_score":71.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2624,"primary_metric":"exact_match","primary_value":0.7166666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":4,"failure_rate":0.06666666666666667,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":806.0470264669372,"latency_p50_ms":733.9355000003707,"latency_p95_ms":1450.6030038486642,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","normalized_quality_score":73.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":2621,"primary_metric":"exact_match","primary_value":0.7333333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":3,"failure_rate":0.05,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1623.2633048988646,"latency_p50_ms":1483.7232709978707,"latency_p95_ms":2773.320496148517,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","normalized_quality_score":73.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2638,"primary_metric":"exact_match","primary_value":0.7333333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","failure_count":2,"failure_rate":0.03333333333333333,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2109.738486816059,"latency_p50_ms":1936.254354484845,"latency_p95_ms":3802.9144431435266,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","normalized_quality_score":73.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2621,"primary_metric":"exact_match","primary_value":0.7333333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-4b-q8-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":2,"failure_rate":0.03333333333333333,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2716.7342707665134,"latency_p50_ms":2522.184229499544,"latency_p95_ms":4643.15792289417,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","normalized_quality_score":76.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2658,"primary_metric":"exact_match","primary_value":0.7666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","failure_count":4,"failure_rate":0.06666666666666667,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2517.9370430715303,"latency_p50_ms":2332.108625007095,"latency_p95_ms":4228.940910479287,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","normalized_quality_score":71.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2696,"primary_metric":"exact_match","primary_value":0.7166666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","failure_count":10,"failure_rate":0.16666666666666666,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":6660.125209748609,"latency_p50_ms":5688.188729494868,"latency_p95_ms":12916.947212500963,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","normalized_quality_score":68.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2395,"primary_metric":"exact_match","primary_value":0.6833333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":1,"failure_rate":0.016666666666666666,"input_tokens_total":8652,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1693.6658090327303,"latency_p50_ms":1434.1119370001252,"latency_p95_ms":3149.145677108754,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","normalized_quality_score":76.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":2690,"primary_metric":"exact_match","primary_value":0.7666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core","sample_count":60}],"dataset_summaries":[{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":6,"failure_rate":0.1,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":4839.719958333929,"latency_p50_ms":4484.091166501457,"latency_p95_ms":7536.193280648876,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","normalized_quality_score":78.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2689,"primary_metric":"exact_match","primary_value":0.7833333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":58,"failure_rate":0.9666666666666667,"input_tokens_total":242,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1515.5912111426005,"latency_p50_ms":160.70060449419543,"latency_p95_ms":2834.6031663066206,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","normalized_quality_score":3.3333333333333335,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":64,"primary_metric":"exact_match","primary_value":0.03333333333333333,"provider_cost_coverage_rate":0.03333333333333333,"provider_cost_known":true,"provider_cost_observed_total_usd":3.735e-05,"provider_cost_per_1k_cases_usd":0.018675,"provider_cost_per_case_usd":1.8675e-05,"provider_cost_priced_cases":2,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":60,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":4986.970465994091,"latency_p50_ms":192.07681200350635,"latency_p95_ms":60002.98497467884,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","normalized_quality_score":0.0,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":null,"primary_metric":"exact_match","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.8-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":10570,"known_provider_cost_quality_pareto":true,"latency_mean_ms":1934.9246188348236,"latency_p50_ms":1730.0426670044544,"latency_p95_ms":2594.3603166953817,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","normalized_quality_score":91.66666666666666,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":3316,"primary_metric":"exact_match","primary_value":0.9166666666666666,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0060932,"provider_cost_per_1k_cases_usd":0.10155333333333333,"provider_cost_per_case_usd":0.00010155333333333333,"provider_cost_priced_cases":60,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":60,"provider_cost_total_usd":0.0060932,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":10570,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2258.3810763889537,"latency_p50_ms":2046.710938011529,"latency_p95_ms":3548.6162735644007,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","normalized_quality_score":83.33333333333334,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":3356,"primary_metric":"exact_match","primary_value":0.8333333333333334,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0061412,"provider_cost_per_1k_cases_usd":0.10235333333333334,"provider_cost_per_case_usd":0.00010235333333333333,"provider_cost_priced_cases":60,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":60,"provider_cost_total_usd":0.0061412,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":8212,"known_provider_cost_quality_pareto":false,"latency_mean_ms":11260.75794856879,"latency_p50_ms":9417.586916533764,"latency_p95_ms":22641.866754152565,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","normalized_quality_score":76.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":35103,"primary_metric":"exact_match","primary_value":0.7666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":2,"failure_rate":0.03333333333333333,"input_tokens_total":8424,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1666.822604850798,"latency_p50_ms":1567.1223544995883,"latency_p95_ms":2798.8641288560757,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","normalized_quality_score":73.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2345,"primary_metric":"exact_match","primary_value":0.7333333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":2,"failure_rate":0.03333333333333333,"input_tokens_total":8424,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2980.6488194495614,"latency_p50_ms":2663.501458497194,"latency_p95_ms":5424.1780169439135,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","normalized_quality_score":71.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2346,"primary_metric":"exact_match","primary_value":0.7166666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":3,"failure_rate":0.05,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":473.98608951771166,"latency_p50_ms":432.87633300496964,"latency_p95_ms":798.6456500082567,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","normalized_quality_score":68.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":2661,"primary_metric":"exact_match","primary_value":0.6833333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":5,"failure_rate":0.08333333333333333,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1611.7719881663409,"latency_p50_ms":1479.928854474565,"latency_p95_ms":3047.5117707916065,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","normalized_quality_score":71.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2624,"primary_metric":"exact_match","primary_value":0.7166666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":4,"failure_rate":0.06666666666666667,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":806.0470264669372,"latency_p50_ms":733.9355000003707,"latency_p95_ms":1450.6030038486642,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","normalized_quality_score":73.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":2621,"primary_metric":"exact_match","primary_value":0.7333333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":3,"failure_rate":0.05,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1623.2633048988646,"latency_p50_ms":1483.7232709978707,"latency_p95_ms":2773.320496148517,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","normalized_quality_score":73.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2638,"primary_metric":"exact_match","primary_value":0.7333333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":2,"failure_rate":0.03333333333333333,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2109.738486816059,"latency_p50_ms":1936.254354484845,"latency_p95_ms":3802.9144431435266,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","normalized_quality_score":73.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2621,"primary_metric":"exact_match","primary_value":0.7333333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":2,"failure_rate":0.03333333333333333,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2716.7342707665134,"latency_p50_ms":2522.184229499544,"latency_p95_ms":4643.15792289417,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","normalized_quality_score":76.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2658,"primary_metric":"exact_match","primary_value":0.7666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":4,"failure_rate":0.06666666666666667,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2517.9370430715303,"latency_p50_ms":2332.108625007095,"latency_p95_ms":4228.940910479287,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","normalized_quality_score":71.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2696,"primary_metric":"exact_match","primary_value":0.7166666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":10,"failure_rate":0.16666666666666666,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":6660.125209748609,"latency_p50_ms":5688.188729494868,"latency_p95_ms":12916.947212500963,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","normalized_quality_score":68.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2395,"primary_metric":"exact_match","primary_value":0.6833333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":1,"failure_rate":0.016666666666666666,"input_tokens_total":8652,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1693.6658090327303,"latency_p50_ms":1434.1119370001252,"latency_p95_ms":3149.145677108754,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","normalized_quality_score":76.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":2690,"primary_metric":"exact_match","primary_value":0.7666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60}]},"disagreements":[],"family_breakdown":[],"schema_version":"2"}},"generated_at_utc":"2026-10-22T14:28:04.247188+00:00","models":{"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a":{"current_cells":[{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T13:18:34.815598+00:00","deployment":"api","effective_model_id":"gpt-6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","experiment_kind":"standard","failure_count":0,"family":"gpt-6","generation_policy":{},"git_commit":"7f796c10d99e934da6417ea54405fbf9352ad399","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.7987012987012987,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"gpt-6-luna-core-gpt-6-luna-intent-classification-core","runtime_key":"openai-api","sample_count":154,"task_id":"intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T15:34:31.093143+00:00","deployment":"api","effective_model_id":"gpt-6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","experiment_kind":"standard","failure_count":0,"family":"gpt-6","generation_policy":{},"git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.7,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","runtime_key":"openai-api","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T13:13:20.819089+00:00","deployment":"api","effective_model_id":"gpt-6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","experiment_kind":"standard","failure_count":0,"family":"gpt-6","generation_policy":{},"git_commit":"7f796c10d99e934da6417ea54405fbf9352ad399","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.425,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core","runtime_key":"openai-api","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T13:25:20.495814+00:00","deployment":"api","effective_model_id":"gpt-6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","experiment_kind":"standard","failure_count":0,"family":"gpt-6","generation_policy":{},"git_commit":"7f796c10d99e934da6417ea54405fbf9352ad399","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.9935064935064936,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"gpt-6-luna-core-gpt-6-luna-oos-calibration-core","runtime_key":"openai-api","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T13:12:07.696210+00:00","deployment":"api","effective_model_id":"gpt-6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","experiment_kind":"standard","failure_count":0,"family":"gpt-6","generation_policy":{},"git_commit":"7f796c10d99e934da6417ea54405fbf9352ad399","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.95,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"gpt-6-luna-core-gpt-6-luna-qa-abstention-core","runtime_key":"openai-api","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T13:10:27.524208+00:00","deployment":"api","effective_model_id":"gpt-6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","experiment_kind":"standard","failure_count":0,"family":"gpt-6","generation_policy":{},"git_commit":"7f796c10d99e934da6417ea54405fbf9352ad399","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.8333333333333334,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"gpt-6-luna-core-gpt-6-luna-structured-output-core","runtime_key":"openai-api","sample_count":60,"task_id":"structured-output"}],"decision_summary":{"artifact_format":null,"artifact_size_bytes":null,"current_run_ids":["f56a81bc-d911-4b76-9b68-361b14e92a2c","gpt-6-luna-core-gpt-6-luna-intent-classification-core","gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core","gpt-6-luna-core-gpt-6-luna-oos-calibration-core","gpt-6-luna-core-gpt-6-luna-qa-abstention-core","gpt-6-luna-core-gpt-6-luna-structured-output-core"],"deployment":"api","effective_model_id":"gpt-6-luna","execution_count":6,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","machine":"arm64","release":"25.6.0","run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":1,"failure_rate":0.001968503937007874,"family":"gpt-6","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":341867,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2153.128683968199,"latency_p50_ms":1896.0361664940137,"latency_p95_ms":3750.8784476871365,"latest_completed_at_utc":"2026-10-15T15:34:31.093143+00:00","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","observed_case_count":508,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":25886,"overall_quality_score":78.3423520923521,"parameters_b":null,"provider_cost_coverage_rate":0.9980314960629921,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0994366,"provider_cost_per_1k_cases_usd":0.19612741617357002,"provider_cost_per_case_usd":0.00019612741617357,"provider_cost_priced_cases":507,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"openai","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":null,"resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":null,"peak_execution_signature":null,"peak_run_id":null,"process_cpu_percent_avg":null,"process_rss_bytes_avg":null,"process_rss_bytes_peak":null,"resource_cell_count":0,"sample_count":0,"sampling_error_count":0,"scope":"unavailable","source":"unavailable","system_available_memory_bytes_min":null},"result_policy":"standard","runtime_key":"openai-api","tags":["api","closed","openai"]},"history":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T13:18:34.815598+00:00","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":0,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.7987012987012987,"profile":"core","result_state":"CURRENT","run_id":"gpt-6-luna-core-gpt-6-luna-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T15:34:31.093143+00:00","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":0,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.7,"profile":"core","result_state":"CURRENT","run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T13:13:20.819089+00:00","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":0,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.425,"profile":"core","result_state":"CURRENT","run_id":"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T13:25:20.495814+00:00","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":0,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.9935064935064936,"profile":"core","result_state":"CURRENT","run_id":"gpt-6-luna-core-gpt-6-luna-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T13:12:07.696210+00:00","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":0,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.95,"profile":"core","result_state":"CURRENT","run_id":"gpt-6-luna-core-gpt-6-luna-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T13:10:27.524208+00:00","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":0,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.8333333333333334,"profile":"core","result_state":"CURRENT","run_id":"gpt-6-luna-core-gpt-6-luna-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"model":{"artifact_format":null,"artifact_size_bytes":null,"completed_at_utc":"2026-10-15T15:34:31.093143+00:00","deployment":"api","effective_model_id":"gpt-6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","family":"gpt-6","git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","parameters_b":null,"provider_key":"openai","quantization":null,"run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","run_status":"COMPLETED","runtime_key":"openai-api"},"model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","schema_version":"2"},"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495":{"current_cells":[{"artifact_format":"gguf","artifact_size_bytes":3775709216,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T11:24:59.796058+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-bf16","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","parameters_b":2.0,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.577922077922078,"profile":"core","provider_key":"korgis","quantization":"BF16","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","model_key":"qwen3.5-2b-bf16","process_cpu_percent_avg":5.488711560059013,"process_rss_bytes_avg":5427256115.200001,"process_rss_bytes_peak":7206486016.0,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core","sample_count":618,"sample_interval_ms":285.9658953851292,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5988106240.0},"run_id":"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":3775709216,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-bf16","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","parameters_b":2.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.25,"profile":"core","provider_key":"korgis","quantization":"BF16","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","model_key":"qwen3.5-2b-bf16","process_cpu_percent_avg":8.718440372188164,"process_rss_bytes_avg":4446060373.333334,"process_rss_bytes_peak":4834885632.0,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":121,"sample_interval_ms":268.43950018325506,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5017370624.0},"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":3775709216,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T11:22:33.733080+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-bf16","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","parameters_b":2.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.375,"profile":"core","provider_key":"korgis","quantization":"BF16","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","model_key":"qwen3.5-2b-bf16","process_cpu_percent_avg":6.761276471009742,"process_rss_bytes_avg":5075235362.133334,"process_rss_bytes_peak":5976489984.0,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core","sample_count":120,"sample_interval_ms":240.70379737386247,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":10141270016.0},"run_id":"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":3775709216,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T11:28:16.241563+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-bf16","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","parameters_b":2.0,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.5,"profile":"core","provider_key":"korgis","quantization":"BF16","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","model_key":"qwen3.5-2b-bf16","process_cpu_percent_avg":5.117048169336818,"process_rss_bytes_avg":5994336968.810392,"process_rss_bytes_peak":9291366400.0,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","sample_count":770,"sample_interval_ms":293.5350881957497,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5950177280.0},"run_id":"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":3775709216,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T11:22:09.705119+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-bf16","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","parameters_b":2.0,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.6333333333333333,"profile":"core","provider_key":"korgis","quantization":"BF16","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","model_key":"qwen3.5-2b-bf16","process_cpu_percent_avg":5.5345916105632815,"process_rss_bytes_avg":5136314017.564444,"process_rss_bytes_peak":5894602752.0,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core","sample_count":226,"sample_interval_ms":279.53779863981055,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":7390167040.0},"run_id":"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":3775709216,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T11:21:16.650276+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-bf16","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","parameters_b":2.0,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7166666666666667,"profile":"core","provider_key":"korgis","quantization":"BF16","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","model_key":"qwen3.5-2b-bf16","process_cpu_percent_avg":5.386938078405328,"process_rss_bytes_avg":4179186975.941509,"process_rss_bytes_peak":4437606400.0,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core","sample_count":391,"sample_interval_ms":306.911332594406,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4923883520.0},"run_id":"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"}],"decision_summary":{"artifact_format":"gguf","artifact_size_bytes":3775709216,"current_run_ids":["b10ad834-2bdd-409b-9073-49f0233792ea","qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core","qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core","qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core","qwen-variants-core-qwen3.5-2b-bf16-structured-output-core"],"deployment":"local","effective_model_id":"qwen3.5-2b-bf16","execution_count":6,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","machine":"arm64","release":"25.6.0","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":15,"failure_rate":0.02952755905511811,"family":"qwen3.5","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":204788,"known_provider_cost_quality_pareto":false,"latency_mean_ms":902.998680881693,"latency_p50_ms":822.7691460051574,"latency_p95_ms":1520.380753971403,"latest_completed_at_utc":"2026-10-16T08:53:37.155498+00:00","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","observed_case_count":508,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":8763,"overall_quality_score":50.88203463203464,"parameters_b":2.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":508,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"BF16","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","machine":"arm64","release":"25.6.0","run_id":"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","peak_run_id":"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","process_cpu_percent_avg":6.167834376927058,"process_rss_bytes_avg":5043064968.8305025,"process_rss_bytes_peak":9291366400.0,"resource_cell_count":6,"sample_count":2246,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4923883520.0},"result_policy":"standard","runtime_key":"korgis-local","tags":["local","open-weight","qwen","small","bf16"]},"history":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T11:24:59.796058+00:00","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","failure_count":0,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.577922077922078,"profile":"core","result_state":"CURRENT","run_id":"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","failure_count":0,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.25,"profile":"core","result_state":"CURRENT","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T11:22:33.733080+00:00","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","failure_count":0,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.375,"profile":"core","result_state":"CURRENT","run_id":"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T11:28:16.241563+00:00","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","failure_count":0,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.5,"profile":"core","result_state":"CURRENT","run_id":"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T11:22:09.705119+00:00","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","failure_count":0,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.6333333333333333,"profile":"core","result_state":"CURRENT","run_id":"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T11:21:16.650276+00:00","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","failure_count":0,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7166666666666667,"profile":"core","result_state":"CURRENT","run_id":"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"model":{"artifact_format":"gguf","artifact_size_bytes":3775709216,"completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-bf16","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","family":"qwen3.5","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","parameters_b":2.0,"provider_key":"korgis","quantization":"BF16","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","run_status":"PARTIAL","runtime_key":"korgis-local"},"model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","schema_version":"2"},"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6":{"current_cells":[{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-06T12:50:59.652242+00:00","deployment":"api","effective_model_id":"jev-1.13.0","execution_signature":"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3","experiment_kind":"standard","failure_count":0,"family":"jev","generation_policy":{},"git_commit":"72697127df45f8a3abc98b0b7bdf6fd53a0b5c98","model_id":"jev-1.13.0","model_key":"jev-1.13.0","model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.7467532467532467,"profile":"core","provider_key":"typesafe","quantization":null,"resource_summary":null,"run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","runtime_key":"typesafe-api","sample_count":154,"task_id":"intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-06T12:51:50.754874+00:00","deployment":"api","effective_model_id":"jev-1.13.0","execution_signature":"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3","experiment_kind":"standard","failure_count":0,"family":"jev","generation_policy":{},"git_commit":"72697127df45f8a3abc98b0b7bdf6fd53a0b5c98","model_id":"jev-1.13.0","model_key":"jev-1.13.0","model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.974025974025974,"profile":"core","provider_key":"typesafe","quantization":null,"resource_summary":null,"run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","runtime_key":"typesafe-api","sample_count":154,"task_id":"calibrated-intent-classification"}],"decision_summary":{"artifact_format":null,"artifact_size_bytes":null,"current_run_ids":["2026-10-core-v1-jev-1.13.0-intent-classification-core","2026-10-core-v1-jev-1.13.0-oos-calibration-core"],"deployment":"api","effective_model_id":"jev-1.13.0","execution_count":2,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3","machine":"arm64","release":"25.6.0","run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":0,"failure_rate":0.0,"family":"jev","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":1062302,"known_provider_cost_quality_pareto":true,"latency_mean_ms":258.10247596585555,"latency_p50_ms":248.95504149753833,"latency_p95_ms":315.9772399922076,"latest_completed_at_utc":"2026-10-06T12:51:50.754874+00:00","model_id":"jev-1.13.0","model_key":"jev-1.13.0","model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","observed_case_count":616,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":true,"observed_quality_parameters_pareto":false,"output_tokens_total":511838,"overall_quality_score":86.03896103896105,"parameters_b":null,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.044616684000000004,"provider_cost_per_1k_cases_usd":0.07242968181818182,"provider_cost_per_case_usd":7.242968181818182e-05,"provider_cost_priced_cases":616,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"jev","processing":"standard","source":"TypeSafe AI public pricing: $42 per billion input tokens; output free.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":616,"provider_cost_total_usd":0.044616684000000004,"provider_key":"typesafe","quality_coverage_capabilities":2,"quality_coverage_complete":false,"quality_coverage_datasets":2,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":null,"resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":null,"peak_execution_signature":null,"peak_run_id":null,"process_cpu_percent_avg":null,"process_rss_bytes_avg":null,"process_rss_bytes_peak":null,"resource_cell_count":0,"sample_count":0,"sampling_error_count":0,"scope":"unavailable","source":"unavailable","system_available_memory_bytes_min":null},"result_policy":"standard","runtime_key":"typesafe-api","tags":["api","closed","typesafe","decision-model"]},"history":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-06T12:50:59.652242+00:00","execution_signature":"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3","failure_count":0,"model_key":"jev-1.13.0","model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.7467532467532467,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-06T12:51:50.754874+00:00","execution_signature":"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3","failure_count":0,"model_key":"jev-1.13.0","model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.974025974025974,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"}],"model":{"artifact_format":null,"artifact_size_bytes":null,"completed_at_utc":"2026-10-06T12:51:50.754874+00:00","deployment":"api","effective_model_id":"jev-1.13.0","execution_signature":"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3","family":"jev","git_commit":"72697127df45f8a3abc98b0b7bdf6fd53a0b5c98","model_id":"jev-1.13.0","model_key":"jev-1.13.0","model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","parameters_b":null,"provider_key":"typesafe","quantization":null,"run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","run_status":"COMPLETED","runtime_key":"typesafe-api"},"model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","schema_version":"2"},"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a":{"current_cells":[{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T13:59:58.574010+00:00","deployment":"api","effective_model_id":"gemini-3.5-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.5","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.0,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core","runtime_key":"gemini-api","sample_count":154,"task_id":"intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"api","effective_model_id":"gemini-3.5-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.5","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.125,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"gemini-api","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T13:59:36.375684+00:00","deployment":"api","effective_model_id":"gemini-3.5-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.5","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.0,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core","runtime_key":"gemini-api","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T14:00:21.676846+00:00","deployment":"api","effective_model_id":"gemini-3.5-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.5","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.5,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core","runtime_key":"gemini-api","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T13:59:29.849450+00:00","deployment":"api","effective_model_id":"gemini-3.5-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.5","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.016666666666666666,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core","runtime_key":"gemini-api","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T13:59:13.872786+00:00","deployment":"api","effective_model_id":"gemini-3.5-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.5","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.03333333333333333,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core","runtime_key":"gemini-api","sample_count":60,"task_id":"structured-output"}],"decision_summary":{"artifact_format":null,"artifact_size_bytes":null,"current_run_ids":["b10ad834-2bdd-409b-9073-49f0233792ea","gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core","gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core","gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core","gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core","gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core"],"deployment":"api","effective_model_id":"gemini-3.5-flash","execution_count":6,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","machine":"arm64","release":"25.6.0","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":500,"failure_rate":0.984251968503937,"family":"gemini-3.5","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":791,"known_provider_cost_quality_pareto":false,"latency_mean_ms":334.08824837785835,"latency_p50_ms":133.2621249894146,"latency_p95_ms":213.5791892651468,"latest_completed_at_utc":"2026-10-16T08:53:37.155498+00:00","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","observed_case_count":508,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":true,"observed_quality_parameters_pareto":false,"output_tokens_total":113,"overall_quality_score":11.25,"parameters_b":null,"provider_cost_coverage_rate":0.015748031496062992,"provider_cost_known":true,"provider_cost_observed_total_usd":9.3225e-05,"provider_cost_per_1k_cases_usd":0.011653125,"provider_cost_per_case_usd":1.1653125e-05,"provider_cost_priced_cases":8,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"gemini","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":null,"resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":null,"peak_execution_signature":null,"peak_run_id":null,"process_cpu_percent_avg":null,"process_rss_bytes_avg":null,"process_rss_bytes_peak":null,"resource_cell_count":0,"sample_count":0,"sampling_error_count":0,"scope":"unavailable","source":"unavailable","system_available_memory_bytes_min":null},"result_policy":"standard","runtime_key":"gemini-api","tags":["api","closed","google","gemini","flash"]},"history":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T13:59:58.574010+00:00","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":0,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.0,"profile":"core","result_state":"CURRENT","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":0,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.125,"profile":"core","result_state":"CURRENT","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T13:59:36.375684+00:00","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":0,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.0,"profile":"core","result_state":"CURRENT","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T14:00:21.676846+00:00","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":0,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.5,"profile":"core","result_state":"CURRENT","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T13:59:29.849450+00:00","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":0,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.016666666666666666,"profile":"core","result_state":"CURRENT","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T13:59:13.872786+00:00","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":0,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.03333333333333333,"profile":"core","result_state":"CURRENT","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"model":{"artifact_format":null,"artifact_size_bytes":null,"completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"api","effective_model_id":"gemini-3.5-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","family":"gemini-3.5","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","parameters_b":null,"provider_key":"gemini","quantization":null,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","run_status":"PARTIAL","runtime_key":"gemini-api"},"model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","schema_version":"2"},"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65":{"current_cells":[{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:382be54f0e8c250212fa1d53656a34eb22335c491d7aa5b1ba59e927a0c58dea","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T14:59:31.170321+00:00","deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","experiment_kind":"standard","failure_count":0,"family":"limite","generation_policy":{},"git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65","parameters_b":1.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.7222222222222222,"profile":"extended","provider_key":"transformers","quantization":null,"resource_summary":null,"run_id":"2533bf73-b012-4d12-a6b6-9b6e4334dae2","runtime_key":"transformers-local","sample_count":72,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:bff2038c9a70a239e320793c0c579d02b35645dd0f722e6fd8e0c6a9b30b4b07","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T13:35:21.786434+00:00","deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","experiment_kind":"standard","failure_count":0,"family":"limite","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65","parameters_b":1.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.8333333333333334,"profile":"smoke","provider_key":"transformers","quantization":null,"resource_summary":null,"run_id":"cb837fcf-570f-4a89-bbbd-cded7a3c335b","runtime_key":"transformers-local","sample_count":12,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T13:46:02.371339+00:00","deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","experiment_kind":"standard","failure_count":0,"family":"limite","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65","parameters_b":1.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.725,"profile":"core","provider_key":"transformers","quantization":null,"resource_summary":null,"run_id":"954053b1-6f87-4196-bb2a-028404b6fef1","runtime_key":"transformers-local","sample_count":40,"task_id":"mathematical-reasoning"}],"decision_summary":null,"history":[{"benchmark_signature":"sha256:benchmark:382be54f0e8c250212fa1d53656a34eb22335c491d7aa5b1ba59e927a0c58dea","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T14:59:31.170321+00:00","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","failure_count":0,"model_key":"limite-1b-violetto","model_signature":"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.7222222222222222,"profile":"extended","result_state":"CURRENT","run_id":"2533bf73-b012-4d12-a6b6-9b6e4334dae2","sample_count":72,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:bff2038c9a70a239e320793c0c579d02b35645dd0f722e6fd8e0c6a9b30b4b07","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T13:35:21.786434+00:00","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","failure_count":0,"model_key":"limite-1b-violetto","model_signature":"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.8333333333333334,"profile":"smoke","result_state":"CURRENT","run_id":"cb837fcf-570f-4a89-bbbd-cded7a3c335b","sample_count":12,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T13:46:02.371339+00:00","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","failure_count":0,"model_key":"limite-1b-violetto","model_signature":"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.725,"profile":"core","result_state":"CURRENT","run_id":"954053b1-6f87-4196-bb2a-028404b6fef1","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"model":{"artifact_format":null,"artifact_size_bytes":null,"completed_at_utc":"2026-10-15T14:59:31.170321+00:00","deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","family":"limite","git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65","parameters_b":1.0,"provider_key":"transformers","quantization":null,"run_id":"2533bf73-b012-4d12-a6b6-9b6e4334dae2","run_status":"COMPLETED","runtime_key":"transformers-local"},"model_signature":"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65","schema_version":"2"},"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69":{"current_cells":[{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:27:15.751735+00:00","deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","experiment_kind":"standard","failure_count":0,"family":"limite","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","parameters_b":1.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.9,"profile":"core","provider_key":"transformers","quantization":null,"resource_summary":null,"run_id":"16a842b0-5dcc-47b5-8808-850b2ad1ac10","runtime_key":"transformers-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-17T21:24:56.752644+00:00","deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","experiment_kind":"standard","failure_count":0,"family":"limite","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","parameters_b":1.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.8,"profile":"core","provider_key":"transformers","quantization":null,"resource_summary":null,"run_id":"8eef4bab-617a-4638-9e51-53e72034e821","runtime_key":"transformers-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:1fa41ef05579230896c951f4e01da57ccf719647ad6f8235008243556409a7cb","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-20T10:03:44.604674+00:00","deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","experiment_kind":"standard","failure_count":0,"family":"limite","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","parameters_b":1.0,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.0,"profile":"smoke","provider_key":"transformers","quantization":null,"resource_summary":null,"run_id":"b6b4123a-1d20-4c39-8644-ea00c1e5893b","runtime_key":"transformers-local","sample_count":12,"task_id":"qa-abstention"}],"decision_summary":{"artifact_format":null,"artifact_size_bytes":null,"current_run_ids":["16a842b0-5dcc-47b5-8808-850b2ad1ac10","8eef4bab-617a-4638-9e51-53e72034e821","b6b4123a-1d20-4c39-8644-ea00c1e5893b"],"deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_count":3,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","machine":"arm64","release":"25.6.0","run_id":"b6b4123a-1d20-4c39-8644-ea00c1e5893b","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"smoke","failure_count":14,"failure_rate":0.15217391304347827,"family":"limite","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":5563,"known_provider_cost_quality_pareto":false,"latency_mean_ms":19307.593355981542,"latency_p50_ms":14347.055187972728,"latency_p95_ms":51331.97907336871,"latest_completed_at_utc":"2026-10-20T10:03:44.604674+00:00","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","observed_case_count":92,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":true,"output_tokens_total":71678,"overall_quality_score":56.666666666666664,"parameters_b":1.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":92,"provider_cost_total_usd":null,"provider_key":"transformers","quality_coverage_capabilities":3,"quality_coverage_complete":false,"quality_coverage_datasets":3,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":null,"resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":null,"peak_execution_signature":null,"peak_run_id":null,"process_cpu_percent_avg":null,"process_rss_bytes_avg":null,"process_rss_bytes_peak":null,"resource_cell_count":0,"sample_count":0,"sampling_error_count":0,"scope":"unavailable","source":"unavailable","system_available_memory_bytes_min":null},"result_policy":"standard","runtime_key":"transformers-local","tags":["local","open-weight","limite","violetto","mathematics","reasoning","small","1b","in-process"]},"history":[{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:27:15.751735+00:00","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","failure_count":0,"model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.9,"profile":"core","result_state":"CURRENT","run_id":"16a842b0-5dcc-47b5-8808-850b2ad1ac10","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:42:51.527523+00:00","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","failure_count":0,"model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.8,"profile":"core","result_state":"HISTORICAL","run_id":"e0c2dc7c-ff6a-4895-bb32-54ec8a2448d6","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-17T21:24:56.752644+00:00","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","failure_count":0,"model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.8,"profile":"core","result_state":"CURRENT","run_id":"8eef4bab-617a-4638-9e51-53e72034e821","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:1fa41ef05579230896c951f4e01da57ccf719647ad6f8235008243556409a7cb","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-20T10:03:44.604674+00:00","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","failure_count":0,"model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.0,"profile":"smoke","result_state":"CURRENT","run_id":"b6b4123a-1d20-4c39-8644-ea00c1e5893b","sample_count":12,"status":"COMPLETED","task_id":"qa-abstention"}],"model":{"artifact_format":null,"artifact_size_bytes":null,"completed_at_utc":"2026-10-20T10:03:44.604674+00:00","deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","family":"limite","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","parameters_b":1.0,"provider_key":"transformers","quantization":null,"run_id":"b6b4123a-1d20-4c39-8644-ea00c1e5893b","run_status":"COMPLETED","runtime_key":"transformers-local"},"model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","schema_version":"2"},"sha256:model:3fe1bfff5bb2c952f96bb799970eb76d9db17f9f880bc3326f0bdfbdd51577cc":{"current_cells":[{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T12:48:39.598492+00:00","deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:ff14676ac39e23af360cd1209491ef1c14e3dab9a3198b48054de5c6de699fdb","experiment_kind":"standard","failure_count":0,"family":"limite","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:3fe1bfff5bb2c952f96bb799970eb76d9db17f9f880bc3326f0bdfbdd51577cc","parameters_b":1.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.0,"profile":"core","provider_key":"vllm","quantization":null,"resource_summary":null,"run_id":"5623e5c1-213d-447a-88e9-ac51a69a92fa","runtime_key":"vllm-local","sample_count":40,"task_id":"mathematical-reasoning"}],"decision_summary":null,"history":[{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T12:48:39.598492+00:00","execution_signature":"sha256:execution:ff14676ac39e23af360cd1209491ef1c14e3dab9a3198b48054de5c6de699fdb","failure_count":0,"model_key":"limite-1b-violetto","model_signature":"sha256:model:3fe1bfff5bb2c952f96bb799970eb76d9db17f9f880bc3326f0bdfbdd51577cc","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.0,"profile":"core","result_state":"CURRENT","run_id":"5623e5c1-213d-447a-88e9-ac51a69a92fa","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"model":{"artifact_format":null,"artifact_size_bytes":null,"completed_at_utc":"2026-10-15T12:48:39.598492+00:00","deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:ff14676ac39e23af360cd1209491ef1c14e3dab9a3198b48054de5c6de699fdb","family":"limite","git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:3fe1bfff5bb2c952f96bb799970eb76d9db17f9f880bc3326f0bdfbdd51577cc","parameters_b":1.0,"provider_key":"vllm","quantization":null,"run_id":"5623e5c1-213d-447a-88e9-ac51a69a92fa","run_status":"COMPLETED","runtime_key":"vllm-local"},"model_signature":"sha256:model:3fe1bfff5bb2c952f96bb799970eb76d9db17f9f880bc3326f0bdfbdd51577cc","schema_version":"2"},"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481":{"current_cells":[{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T16:36:39.120802+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"cb061291446b8dbd79f524a4fefcd4a4406e2a0d","model_id":"unsloth/qwen3.5-4b","model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6688311688311688,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-4b-q4km","process_cpu_percent_avg":6.008969494809468,"process_rss_bytes_avg":5011243553.1201,"process_rss_bytes_peak":7911129088.0,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":1067,"sample_interval_ms":281.19401557671836,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5343313920.0},"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":2740937888,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:d6ef7d6b6d33f70d69c5d1df4568941686a610cda0bde994801472a902500dd4","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"unsloth/qwen3.5-4b","model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","parameters_b":4.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.4,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:d6ef7d6b6d33f70d69c5d1df4568941686a610cda0bde994801472a902500dd4","model_key":"qwen3.5-4b-q4km","process_cpu_percent_avg":10.67374362291546,"process_rss_bytes_avg":4568710143.999999,"process_rss_bytes_peak":5363400704.0,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":129,"sample_interval_ms":303.8685138652605,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5199380480.0},"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T16:36:39.120802+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"cb061291446b8dbd79f524a4fefcd4a4406e2a0d","model_id":"unsloth/qwen3.5-4b","model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.475,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-4b-q4km","process_cpu_percent_avg":12.689180073192913,"process_rss_bytes_avg":5970077900.800001,"process_rss_bytes_peak":7198195712.0,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":134,"sample_interval_ms":272.2716644006747,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":8184053760.0},"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T16:36:39.120802+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"cb061291446b8dbd79f524a4fefcd4a4406e2a0d","model_id":"unsloth/qwen3.5-4b","model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.974025974025974,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-4b-q4km","process_cpu_percent_avg":6.406836413976952,"process_rss_bytes_avg":6021487127.958771,"process_rss_bytes_peak":10610851840.0,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":1193,"sample_interval_ms":286.7970046134617,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5619220480.0},"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T16:36:39.120802+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"cb061291446b8dbd79f524a4fefcd4a4406e2a0d","model_id":"unsloth/qwen3.5-4b","model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-4b-q4km","process_cpu_percent_avg":9.78397589283007,"process_rss_bytes_avg":7517191058.123175,"process_rss_bytes_peak":10744905728.0,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":261,"sample_interval_ms":276.8212083626713,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6344114176.0},"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T16:36:39.120802+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"cb061291446b8dbd79f524a4fefcd4a4406e2a0d","model_id":"unsloth/qwen3.5-4b","model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7333333333333333,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-4b-q4km","process_cpu_percent_avg":8.486342485499302,"process_rss_bytes_avg":5121284030.324117,"process_rss_bytes_peak":6889406464.0,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":411,"sample_interval_ms":286.355363495796,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5873369088.0},"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"}],"decision_summary":{"artifact_format":"gguf","artifact_size_bytes":null,"current_run_ids":["3a46714c-d579-4688-a8b0-558183b4e02a","b10ad834-2bdd-409b-9073-49f0233792ea"],"deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_count":2,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:d6ef7d6b6d33f70d69c5d1df4568941686a610cda0bde994801472a902500dd4","machine":"arm64","release":"25.6.0","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":3,"failure_rate":0.005905511811023622,"family":"qwen3.5","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":204788,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1429.9514306818367,"latency_p50_ms":1585.6053334937315,"latency_p95_ms":1972.186656796839,"latest_completed_at_utc":"2026-10-16T08:53:37.155498+00:00","model_id":"unsloth/qwen3.5-4b","model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","observed_case_count":508,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":8520,"overall_quality_score":68.63095238095238,"parameters_b":4,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":508,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","machine":"arm64","release":"25.6.0","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","peak_run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","process_cpu_percent_avg":9.008174663870694,"process_rss_bytes_avg":5701665635.721027,"process_rss_bytes_peak":10744905728.0,"resource_cell_count":6,"sample_count":3195,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5199380480.0},"result_policy":"standard","runtime_key":"korgis-local","tags":["local","open-weight","qwen","4b"]},"history":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T16:36:39.120802+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6688311688311688,"profile":"core","result_state":"CURRENT","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","execution_signature":"sha256:execution:d6ef7d6b6d33f70d69c5d1df4568941686a610cda0bde994801472a902500dd4","failure_count":0,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.4,"profile":"core","result_state":"CURRENT","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T16:36:39.120802+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.475,"profile":"core","result_state":"CURRENT","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T16:36:39.120802+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.974025974025974,"profile":"core","result_state":"CURRENT","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T16:36:39.120802+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"profile":"core","result_state":"CURRENT","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T16:36:39.120802+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7333333333333333,"profile":"core","result_state":"CURRENT","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"model":{"artifact_format":"gguf","artifact_size_bytes":2740937888,"completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:d6ef7d6b6d33f70d69c5d1df4568941686a610cda0bde994801472a902500dd4","family":"qwen3.5","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"unsloth/qwen3.5-4b","model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","parameters_b":4.0,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","run_status":"PARTIAL","runtime_key":"korgis-local"},"model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","schema_version":"2"},"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34":{"current_cells":[{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T12:03:20.624151+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.5584415584415584,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b","process_cpu_percent_avg":5.91751681481258,"process_rss_bytes_avg":5777497087.999999,"process_rss_bytes_peak":10717691904.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-intent-classification-core","sample_count":1252,"sample_interval_ms":293.3800480326384,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6120767488.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":2837072896,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T15:45:29.838316+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b","execution_signature":"sha256:execution:0caf2fff221781464037ba78ff0ed5a8158d632ab0e66481fc8df733dd8785c6","experiment_kind":"standard","failure_count":0,"family":"nemotron-nano","generation_policy":{},"git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","parameters_b":4.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.5,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:0caf2fff221781464037ba78ff0ed5a8158d632ab0e66481fc8df733dd8785c6","model_key":"nemotron-nano-4b","process_cpu_percent_avg":10.014690415304136,"process_rss_bytes_avg":4240616618.6666665,"process_rss_bytes_peak":4889542656.0,"run_id":"3b0292d4-8f8f-4843-900c-946c32476964","sample_count":153,"sample_interval_ms":267.2538513759112,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5000953856.0},"run_id":"3b0292d4-8f8f-4843-900c-946c32476964","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T11:57:43.663487+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.575,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b","process_cpu_percent_avg":11.038100201393934,"process_rss_bytes_avg":7133634082.133334,"process_rss_bytes_peak":11070996480.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core","sample_count":143,"sample_interval_ms":255.84691734608595,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":11042144256.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T12:10:00.371141+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b","process_cpu_percent_avg":5.490567532375733,"process_rss_bytes_avg":5130473995.082253,"process_rss_bytes_peak":10141679616.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core","sample_count":1444,"sample_interval_ms":296.66646442619066,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6067978240.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T11:57:09.469183+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b","process_cpu_percent_avg":9.533096212147692,"process_rss_bytes_avg":7103316869.12,"process_rss_bytes_peak":11089903616.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","sample_count":264,"sample_interval_ms":273.58964924757555,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6888439808.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T11:56:03.800158+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7333333333333333,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b","process_cpu_percent_avg":8.190733436222855,"process_rss_bytes_avg":4822041698.065609,"process_rss_bytes_peak":6531989504.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-structured-output-core","sample_count":423,"sample_interval_ms":283.9724207012575,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5921701888.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"}],"decision_summary":{"artifact_format":"gguf","artifact_size_bytes":null,"current_run_ids":["2026-10-core-v1-nemotron-nano-4b-intent-classification-core","2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core","2026-10-core-v1-nemotron-nano-4b-oos-calibration-core","2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","2026-10-core-v1-nemotron-nano-4b-structured-output-core","3b0292d4-8f8f-4843-900c-946c32476964"],"deployment":"local","effective_model_id":"nemotron-nano-4b","execution_count":6,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:0caf2fff221781464037ba78ff0ed5a8158d632ab0e66481fc8df733dd8785c6","machine":"arm64","release":"25.6.0","run_id":"3b0292d4-8f8f-4843-900c-946c32476964","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":2,"failure_rate":0.003937007874015748,"family":"nemotron-nano","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":228654,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1742.2881335475013,"latency_p50_ms":1999.5908125056303,"latency_p95_ms":2575.8644812565763,"latest_completed_at_utc":"2026-10-15T15:45:29.838316+00:00","model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","observed_case_count":508,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":true,"output_tokens_total":8779,"overall_quality_score":71.57467532467533,"parameters_b":4,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":508,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","machine":"arm64","release":"25.6.0","run_id":"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","peak_run_id":"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","process_cpu_percent_avg":8.364117435376155,"process_rss_bytes_avg":5701263391.844644,"process_rss_bytes_peak":11089903616.0,"resource_cell_count":6,"sample_count":3679,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5000953856.0},"result_policy":"standard","runtime_key":"korgis-local","tags":["local","open-weight","nemotron","small"]},"history":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T12:03:20.624151+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.5584415584415584,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-nemotron-nano-4b-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T15:45:29.838316+00:00","execution_signature":"sha256:execution:0caf2fff221781464037ba78ff0ed5a8158d632ab0e66481fc8df733dd8785c6","failure_count":0,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.5,"profile":"core","result_state":"CURRENT","run_id":"3b0292d4-8f8f-4843-900c-946c32476964","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T11:57:43.663487+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.575,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T12:10:00.371141+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T11:57:09.469183+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T11:56:03.800158+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7333333333333333,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-nemotron-nano-4b-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"model":{"artifact_format":"gguf","artifact_size_bytes":2837072896,"completed_at_utc":"2026-10-15T15:45:29.838316+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b","execution_signature":"sha256:execution:0caf2fff221781464037ba78ff0ed5a8158d632ab0e66481fc8df733dd8785c6","family":"nemotron-nano","git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","parameters_b":4.0,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"3b0292d4-8f8f-4843-900c-946c32476964","run_status":"COMPLETED","runtime_key":"korgis-local"},"model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","schema_version":"2"},"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2":{"current_cells":[{"artifact_format":"gguf","artifact_size_bytes":4121781472,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T11:55:33.171703+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-ud-q2kxl","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"qwen/qwen3.5-9b-ud-q2kxl","model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","parameters_b":9.0,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6883116883116883,"profile":"core","provider_key":"korgis","quantization":"UD-Q2_K_XL","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","model_key":"qwen3.5-9b-ud-q2kxl","process_cpu_percent_avg":3.288803542984178,"process_rss_bytes_avg":6954068738.250033,"process_rss_bytes_peak":11224416256.0,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","sample_count":1539,"sample_interval_ms":314.5480896344338,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5547704320.0},"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":4121781472,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T11:48:03.359698+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-ud-q2kxl","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"qwen/qwen3.5-9b-ud-q2kxl","model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","parameters_b":9.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.55,"profile":"core","provider_key":"korgis","quantization":"UD-Q2_K_XL","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","model_key":"qwen3.5-9b-ud-q2kxl","process_cpu_percent_avg":4.913551711729864,"process_rss_bytes_avg":7164752896.0,"process_rss_bytes_peak":9602973696.0,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core","sample_count":164,"sample_interval_ms":288.91353003637056,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":12627738624.0},"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":4121781472,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T12:04:10.873473+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-ud-q2kxl","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"qwen/qwen3.5-9b-ud-q2kxl","model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","parameters_b":9.0,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"profile":"core","provider_key":"korgis","quantization":"UD-Q2_K_XL","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","model_key":"qwen3.5-9b-ud-q2kxl","process_cpu_percent_avg":3.489486960210178,"process_rss_bytes_avg":6127369651.484418,"process_rss_bytes_peak":9636659200.0,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core","sample_count":1713,"sample_interval_ms":322.34991046526653,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5265096704.0},"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":4121781472,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T11:47:21.635679+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-ud-q2kxl","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"qwen/qwen3.5-9b-ud-q2kxl","model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","parameters_b":9.0,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"profile":"core","provider_key":"korgis","quantization":"UD-Q2_K_XL","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","model_key":"qwen3.5-9b-ud-q2kxl","process_cpu_percent_avg":4.725490643337322,"process_rss_bytes_avg":7851091423.1669855,"process_rss_bytes_peak":11102388224.0,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core","sample_count":336,"sample_interval_ms":289.6220233097523,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":8726102016.0},"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":4121781472,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T11:45:53.885678+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-ud-q2kxl","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"qwen/qwen3.5-9b-ud-q2kxl","model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","parameters_b":9.0,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7166666666666667,"profile":"core","provider_key":"korgis","quantization":"UD-Q2_K_XL","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","model_key":"qwen3.5-9b-ud-q2kxl","process_cpu_percent_avg":4.974081455251981,"process_rss_bytes_avg":7070526205.54296,"process_rss_bytes_peak":10778755072.0,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core","sample_count":559,"sample_interval_ms":311.91939930876663,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5600329728.0},"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"}],"decision_summary":{"artifact_format":"gguf","artifact_size_bytes":4121781472,"current_run_ids":["qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core","qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core","qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core","qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core"],"deployment":"local","effective_model_id":"qwen3.5-9b-ud-q2kxl","execution_count":5,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","machine":"arm64","release":"25.6.0","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":4,"failure_rate":0.008547008547008548,"family":"qwen3.5","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":200390,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2500.572187915712,"latency_p50_ms":2746.6233744926285,"latency_p95_ms":3354.4908000039864,"latest_completed_at_utc":"2026-10-13T12:04:10.873473+00:00","model_id":"qwen/qwen3.5-9b-ud-q2kxl","model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","observed_case_count":468,"observed_quality_artifact_size_pareto":true,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":8167,"overall_quality_score":75.65367965367966,"parameters_b":9.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":468,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":468,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":5,"quality_coverage_complete":false,"quality_coverage_datasets":5,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"UD-Q2_K_XL","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","machine":"arm64","release":"25.6.0","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","peak_run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","process_cpu_percent_avg":4.278282862702705,"process_rss_bytes_avg":7033561782.88888,"process_rss_bytes_peak":11224416256.0,"resource_cell_count":5,"sample_count":4311,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5265096704.0},"result_policy":"standard","runtime_key":"korgis-local","tags":["local","open-weight","qwen","9b","q2"]},"history":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T11:55:33.171703+00:00","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","failure_count":0,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6883116883116883,"profile":"core","result_state":"CURRENT","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","execution_signature":"sha256:execution:1b1fd8e53d6820b3449bce86d39dff358b1d6b5726cab7289358ad7888484aa8","failure_count":0,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":null,"profile":"core","result_state":"PARTIAL","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":0,"status":"PARTIAL","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T11:48:03.359698+00:00","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","failure_count":0,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.55,"profile":"core","result_state":"CURRENT","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T12:04:10.873473+00:00","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","failure_count":0,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"profile":"core","result_state":"CURRENT","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T11:47:21.635679+00:00","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","failure_count":0,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"profile":"core","result_state":"CURRENT","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T11:45:53.885678+00:00","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","failure_count":0,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7166666666666667,"profile":"core","result_state":"CURRENT","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"model":{"artifact_format":"gguf","artifact_size_bytes":4121781472,"completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-ud-q2kxl","execution_signature":"sha256:execution:1b1fd8e53d6820b3449bce86d39dff358b1d6b5726cab7289358ad7888484aa8","family":"qwen3.5","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"qwen/qwen3.5-9b-ud-q2kxl","model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","parameters_b":9.0,"provider_key":"korgis","quantization":"UD-Q2_K_XL","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","run_status":"PARTIAL","runtime_key":"korgis-local"},"model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","schema_version":"2"},"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043":{"current_cells":[{"artifact_format":"gguf","artifact_size_bytes":6192222208,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-06T14:37:14.912068+00:00","deployment":"local","effective_model_id":"qwen3.8-27b-iq1s","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","experiment_kind":"standard","failure_count":0,"family":"qwen3.8","generation_policy":{},"git_commit":"640138c3429882a42797d28074f98d3fea98a63b","model_id":"Qwen/Qwen3.8-27B","model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","parameters_b":27.0,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.5844155844155844,"profile":"core","provider_key":"korgis","quantization":"UD-IQ1_S","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","model_key":"qwen3.8-27b-iq1s","process_cpu_percent_avg":3.5068517339810183,"process_rss_bytes_avg":8038410776.168479,"process_rss_bytes_peak":12158550016.0,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","sample_count":4560,"sample_interval_ms":309.607195745226,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5166546944.0},"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":6192222208,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-06T14:37:12.067024+00:00","deployment":"local","effective_model_id":"qwen3.8-27b-iq1s","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","experiment_kind":"standard","failure_count":0,"family":"qwen3.8","generation_policy":{},"git_commit":"640138c3429882a42797d28074f98d3fea98a63b","model_id":"Qwen/Qwen3.8-27B","model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","parameters_b":27.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.25,"profile":"core","provider_key":"korgis","quantization":"UD-IQ1_S","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","model_key":"qwen3.8-27b-iq1s","process_cpu_percent_avg":7.113533739353301,"process_rss_bytes_avg":9138898947.755379,"process_rss_bytes_peak":12174065664.0,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","sample_count":717,"sample_interval_ms":297.72640370809415,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5485953024.0},"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":6192222208,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-06T14:53:01.287222+00:00","deployment":"local","effective_model_id":"qwen3.8-27b-iq1s","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","experiment_kind":"standard","failure_count":0,"family":"qwen3.8","generation_policy":{},"git_commit":"640138c3429882a42797d28074f98d3fea98a63b","model_id":"Qwen/Qwen3.8-27B","model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","parameters_b":27.0,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.8181818181818182,"profile":"core","provider_key":"korgis","quantization":"UD-IQ1_S","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","model_key":"qwen3.8-27b-iq1s","process_cpu_percent_avg":3.3561674118938543,"process_rss_bytes_avg":8080732806.054676,"process_rss_bytes_peak":11618025472.0,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","sample_count":6492,"sample_interval_ms":307.9710546548535,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5336121344.0},"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":6192222208,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-06T14:37:09.228523+00:00","deployment":"local","effective_model_id":"qwen3.8-27b-iq1s","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","experiment_kind":"standard","failure_count":0,"family":"qwen3.8","generation_policy":{},"git_commit":"640138c3429882a42797d28074f98d3fea98a63b","model_id":"Qwen/Qwen3.8-27B","model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","parameters_b":27.0,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8,"profile":"core","provider_key":"korgis","quantization":"UD-IQ1_S","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","model_key":"qwen3.8-27b-iq1s","process_cpu_percent_avg":5.983418081521953,"process_rss_bytes_avg":9550237266.862207,"process_rss_bytes_peak":13159825408.0,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","sample_count":792,"sample_interval_ms":301.02958077118194,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5182652416.0},"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":6192222208,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-06T14:37:06.404488+00:00","deployment":"local","effective_model_id":"qwen3.8-27b-iq1s","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","experiment_kind":"standard","failure_count":0,"family":"qwen3.8","generation_policy":{},"git_commit":"640138c3429882a42797d28074f98d3fea98a63b","model_id":"Qwen/Qwen3.8-27B","model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","parameters_b":27.0,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.6833333333333333,"profile":"core","provider_key":"korgis","quantization":"UD-IQ1_S","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","model_key":"qwen3.8-27b-iq1s","process_cpu_percent_avg":5.429397449534366,"process_rss_bytes_avg":7678658248.931642,"process_rss_bytes_peak":9314566144.0,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","sample_count":1314,"sample_interval_ms":323.2460322738765,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4307910656.0},"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"}],"decision_summary":{"artifact_format":"gguf","artifact_size_bytes":6192222208,"current_run_ids":["2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core"],"deployment":"local","effective_model_id":"qwen3.8-27b-iq1s","execution_count":5,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","machine":"arm64","release":"25.6.0","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":36,"failure_rate":0.07692307692307693,"family":"qwen3.8","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":189885,"known_provider_cost_quality_pareto":false,"latency_mean_ms":8807.17722295277,"latency_p50_ms":8610.94677100482,"latency_p95_ms":12756.54564594878,"latest_completed_at_utc":"2026-10-06T14:53:01.287222+00:00","model_id":"Qwen/Qwen3.8-27B","model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","observed_case_count":468,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":7958,"overall_quality_score":62.71861471861472,"parameters_b":27.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":468,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":468,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":5,"quality_coverage_complete":false,"quality_coverage_datasets":5,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"UD-IQ1_S","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","machine":"arm64","release":"25.6.0","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","peak_run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","process_cpu_percent_avg":5.0778736832568985,"process_rss_bytes_avg":8497387609.154477,"process_rss_bytes_peak":13159825408.0,"resource_cell_count":5,"sample_count":13875,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4307910656.0},"result_policy":"standard","runtime_key":"korgis-local","tags":["local","open-weight","qwen","reasoning","iq1s","27b"]},"history":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-06T14:37:14.912068+00:00","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","failure_count":0,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.5844155844155844,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","execution_signature":"sha256:execution:1b1fd8e53d6820b3449bce86d39dff358b1d6b5726cab7289358ad7888484aa8","failure_count":0,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":null,"profile":"core","result_state":"PARTIAL","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":0,"status":"PARTIAL","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-06T14:37:12.067024+00:00","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","failure_count":0,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.25,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-06T14:53:01.287222+00:00","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","failure_count":0,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.8181818181818182,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-06T14:37:09.228523+00:00","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","failure_count":0,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-06T14:37:06.404488+00:00","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","failure_count":0,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.6833333333333333,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"model":{"artifact_format":"gguf","artifact_size_bytes":null,"completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"qwen3.8-27b-iq1s","execution_signature":"sha256:execution:1b1fd8e53d6820b3449bce86d39dff358b1d6b5726cab7289358ad7888484aa8","family":"qwen3.8","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"Qwen/Qwen3.8-27B","model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","parameters_b":27.0,"provider_key":"korgis","quantization":"UD-IQ1_S","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","run_status":"PARTIAL","runtime_key":"korgis-local"},"model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","schema_version":"2"},"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95":{"current_cells":[{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:718e761e4ba3410c289a5bba91a6474827caaa8da30e644b3df925ba4d16fbfc","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T15:04:59.981803+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"2e7b64ca38a2db66a9d8b36910d2b2ac5da0c7ce","model_id":"unsloth/qwen3.5-2b","model_key":"decisio-qwen3.5-2b-q4km","model_signature":"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.5974025974025974,"profile":"smoke","provider_key":"decisio","quantization":"Q4_K_M","resource_summary":null,"run_id":"16cd6c17-c9de-422d-8583-1a17959bdf6b","runtime_key":"decisio-local","sample_count":77,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T15:26:12.447493+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"2e7b64ca38a2db66a9d8b36910d2b2ac5da0c7ce","model_id":"unsloth/qwen3.5-2b","model_key":"decisio-qwen3.5-2b-q4km","model_signature":"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6038961038961039,"profile":"core","provider_key":"decisio","quantization":"Q4_K_M","resource_summary":null,"run_id":"1e48c427-4e75-4ace-a465-2aec7e2751cd","runtime_key":"decisio-local","sample_count":154,"task_id":"intent-classification"}],"decision_summary":{"artifact_format":"gguf","artifact_size_bytes":null,"current_run_ids":["1e48c427-4e75-4ace-a465-2aec7e2751cd"],"deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_count":1,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","machine":"arm64","release":"25.6.0","run_id":"1e48c427-4e75-4ace-a465-2aec7e2751cd","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":1,"failure_rate":0.006493506493506494,"family":"qwen3.5","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1155.2659180455662,"latency_p50_ms":1149.7726044981391,"latency_p95_ms":1208.5230768483598,"latest_completed_at_utc":"2026-10-05T15:26:12.447493+00:00","model_id":"unsloth/qwen3.5-2b","model_key":"decisio-qwen3.5-2b-q4km","model_signature":"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95","observed_case_count":154,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":true,"output_tokens_total":1203,"overall_quality_score":60.3896103896104,"parameters_b":2,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"provider_key":"decisio","quality_coverage_capabilities":1,"quality_coverage_complete":false,"quality_coverage_datasets":1,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":null,"peak_execution_signature":null,"peak_run_id":null,"process_cpu_percent_avg":null,"process_rss_bytes_avg":null,"process_rss_bytes_peak":null,"resource_cell_count":0,"sample_count":0,"sampling_error_count":0,"scope":"unavailable","source":"unavailable","system_available_memory_bytes_min":null},"result_policy":"standard","runtime_key":"decisio-local","tags":["local","decisio","open-weight","qwen","decision-model"]},"history":[{"benchmark_signature":"sha256:benchmark:718e761e4ba3410c289a5bba91a6474827caaa8da30e644b3df925ba4d16fbfc","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T15:04:59.981803+00:00","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","failure_count":0,"model_key":"decisio-qwen3.5-2b-q4km","model_signature":"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.5974025974025974,"profile":"smoke","result_state":"CURRENT","run_id":"16cd6c17-c9de-422d-8583-1a17959bdf6b","sample_count":77,"status":"COMPLETED","task_id":"intent-classification"},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T15:26:12.447493+00:00","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","failure_count":0,"model_key":"decisio-qwen3.5-2b-q4km","model_signature":"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6038961038961039,"profile":"core","result_state":"CURRENT","run_id":"1e48c427-4e75-4ace-a465-2aec7e2751cd","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"}],"model":{"artifact_format":"gguf","artifact_size_bytes":null,"completed_at_utc":"2026-10-05T15:26:12.447493+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","family":null,"git_commit":"2e7b64ca38a2db66a9d8b36910d2b2ac5da0c7ce","model_id":"unsloth/qwen3.5-2b","model_key":"decisio-qwen3.5-2b-q4km","model_signature":"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95","parameters_b":null,"provider_key":"decisio","quantization":"Q4_K_M","run_id":"1e48c427-4e75-4ace-a465-2aec7e2751cd","run_status":"COMPLETED","runtime_key":"decisio-local"},"model_signature":"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95","schema_version":"2"},"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e":{"current_cells":[{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T13:24:18.004156+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b-q8","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.551948051948052,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b-q8","process_cpu_percent_avg":4.532544993975123,"process_rss_bytes_avg":5542900553.420978,"process_rss_bytes_peak":9062940672.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core","sample_count":1798,"sample_interval_ms":315.6007340511925,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5441241088.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":4233677952,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b-q8","execution_signature":"sha256:execution:89a74e35df26090fe2c6adb16b01e32373fbcdad6966a1b23cc9b88ba35a5170","experiment_kind":"standard","failure_count":0,"family":"nemotron-nano","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","parameters_b":4.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.475,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:89a74e35df26090fe2c6adb16b01e32373fbcdad6966a1b23cc9b88ba35a5170","model_key":"nemotron-nano-4b-q8","process_cpu_percent_avg":8.818030339465103,"process_rss_bytes_avg":7560277067.093333,"process_rss_bytes_peak":10620682240.0,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":152,"sample_interval_ms":260.26064477991895,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":10035691520.0},"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T13:15:22.380707+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b-q8","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.6,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b-q8","process_cpu_percent_avg":8.601215146103618,"process_rss_bytes_avg":7892413781.333334,"process_rss_bytes_peak":11438456832.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","sample_count":158,"sample_interval_ms":252.1124605016667,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":7386234880.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T13:32:50.472664+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b-q8","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b-q8","process_cpu_percent_avg":5.151403915249351,"process_rss_bytes_avg":5654897500.709869,"process_rss_bytes_peak":8370143232.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core","sample_count":1747,"sample_interval_ms":306.91440564565875,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5491818496.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T13:14:43.428857+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b-q8","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b-q8","process_cpu_percent_avg":7.166483605707706,"process_rss_bytes_avg":6463349882.22984,"process_rss_bytes_peak":10168467456.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core","sample_count":359,"sample_interval_ms":297.72169937988014,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5578047488.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T13:13:01.309902+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b-q8","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7166666666666667,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b-q8","process_cpu_percent_avg":4.8684869265121,"process_rss_bytes_avg":5683850551.952017,"process_rss_bytes_peak":6735822848.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core","sample_count":654,"sample_interval_ms":307.429391476698,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5616762880.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"}],"decision_summary":{"artifact_format":"gguf","artifact_size_bytes":null,"current_run_ids":["2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core","2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core","2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core","2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core","b10ad834-2bdd-409b-9073-49f0233792ea"],"deployment":"local","effective_model_id":"nemotron-nano-4b-q8","execution_count":6,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:89a74e35df26090fe2c6adb16b01e32373fbcdad6966a1b23cc9b88ba35a5170","machine":"arm64","release":"25.6.0","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":2,"failure_rate":0.003937007874015748,"family":"nemotron-nano","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":228654,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2567.452957344186,"latency_p50_ms":2537.835583498236,"latency_p95_ms":4611.300510846194,"latest_completed_at_utc":"2026-10-16T08:53:37.155498+00:00","model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","observed_case_count":508,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":8897,"overall_quality_score":71.18867243867244,"parameters_b":4,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":508,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","machine":"arm64","release":"25.6.0","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","peak_run_id":"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","process_cpu_percent_avg":6.5230274878355,"process_rss_bytes_avg":6466281556.123229,"process_rss_bytes_peak":11438456832.0,"resource_cell_count":6,"sample_count":4868,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5441241088.0},"result_policy":"standard","runtime_key":"korgis-local","tags":["local","open-weight","nemotron","medium"]},"history":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T13:24:18.004156+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.551948051948052,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","execution_signature":"sha256:execution:89a74e35df26090fe2c6adb16b01e32373fbcdad6966a1b23cc9b88ba35a5170","failure_count":0,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.475,"profile":"core","result_state":"CURRENT","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T13:15:22.380707+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.6,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T13:32:50.472664+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T13:14:43.428857+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T13:13:01.309902+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7166666666666667,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"model":{"artifact_format":"gguf","artifact_size_bytes":4233677952,"completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b-q8","execution_signature":"sha256:execution:89a74e35df26090fe2c6adb16b01e32373fbcdad6966a1b23cc9b88ba35a5170","family":"nemotron-nano","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","parameters_b":4.0,"provider_key":"korgis","quantization":"Q8_0","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","run_status":"PARTIAL","runtime_key":"korgis-local"},"model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","schema_version":"2"},"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a":{"current_cells":[{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:718e761e4ba3410c289a5bba91a6474827caaa8da30e644b3df925ba4d16fbfc","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T15:11:29.046506+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"2e7b64ca38a2db66a9d8b36910d2b2ac5da0c7ce","model_id":"unsloth/qwen3.5-4b","model_key":"decisio-qwen3.5-4b-q4km","model_signature":"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6753246753246753,"profile":"smoke","provider_key":"decisio","quantization":"Q4_K_M","resource_summary":null,"run_id":"a97505cf-abc6-4782-bf99-38de2a4492b7","runtime_key":"decisio-local","sample_count":77,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T15:33:33.988827+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"2e7b64ca38a2db66a9d8b36910d2b2ac5da0c7ce","model_id":"unsloth/qwen3.5-4b","model_key":"decisio-qwen3.5-4b-q4km","model_signature":"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.7012987012987013,"profile":"core","provider_key":"decisio","quantization":"Q4_K_M","resource_summary":null,"run_id":"b906e312-aa49-4397-b60e-a1c3ad6a8f87","runtime_key":"decisio-local","sample_count":154,"task_id":"intent-classification"}],"decision_summary":{"artifact_format":"gguf","artifact_size_bytes":null,"current_run_ids":["b906e312-aa49-4397-b60e-a1c3ad6a8f87"],"deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_count":1,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","machine":"arm64","release":"25.6.0","run_id":"b906e312-aa49-4397-b60e-a1c3ad6a8f87","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":1,"failure_rate":0.006493506493506494,"family":"qwen3.5","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2834.3784266303433,"latency_p50_ms":2828.9587710023625,"latency_p95_ms":2931.071354152664,"latest_completed_at_utc":"2026-10-05T15:33:33.988827+00:00","model_id":"unsloth/qwen3.5-4b","model_key":"decisio-qwen3.5-4b-q4km","model_signature":"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a","observed_case_count":154,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":1192,"overall_quality_score":70.12987012987013,"parameters_b":4,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"provider_key":"decisio","quality_coverage_capabilities":1,"quality_coverage_complete":false,"quality_coverage_datasets":1,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":null,"peak_execution_signature":null,"peak_run_id":null,"process_cpu_percent_avg":null,"process_rss_bytes_avg":null,"process_rss_bytes_peak":null,"resource_cell_count":0,"sample_count":0,"sampling_error_count":0,"scope":"unavailable","source":"unavailable","system_available_memory_bytes_min":null},"result_policy":"standard","runtime_key":"decisio-local","tags":["local","decisio","open-weight","qwen","decision-model"]},"history":[{"benchmark_signature":"sha256:benchmark:718e761e4ba3410c289a5bba91a6474827caaa8da30e644b3df925ba4d16fbfc","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T15:11:29.046506+00:00","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","failure_count":0,"model_key":"decisio-qwen3.5-4b-q4km","model_signature":"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6753246753246753,"profile":"smoke","result_state":"CURRENT","run_id":"a97505cf-abc6-4782-bf99-38de2a4492b7","sample_count":77,"status":"COMPLETED","task_id":"intent-classification"},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T15:33:33.988827+00:00","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","failure_count":0,"model_key":"decisio-qwen3.5-4b-q4km","model_signature":"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.7012987012987013,"profile":"core","result_state":"CURRENT","run_id":"b906e312-aa49-4397-b60e-a1c3ad6a8f87","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"}],"model":{"artifact_format":"gguf","artifact_size_bytes":null,"completed_at_utc":"2026-10-05T15:33:33.988827+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","family":null,"git_commit":"2e7b64ca38a2db66a9d8b36910d2b2ac5da0c7ce","model_id":"unsloth/qwen3.5-4b","model_key":"decisio-qwen3.5-4b-q4km","model_signature":"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a","parameters_b":null,"provider_key":"decisio","quantization":"Q4_K_M","run_id":"b906e312-aa49-4397-b60e-a1c3ad6a8f87","run_status":"COMPLETED","runtime_key":"decisio-local"},"model_signature":"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a","schema_version":"2"},"sha256:model:8dcf43803761fd5570f637e166b655565b6c638e51a6a96848f58f6e6699a734":{"current_cells":[{"artifact_format":"gguf","artifact_size_bytes":2600224352,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T15:45:57.052259+00:00","deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_signature":"sha256:execution:d26122294632fb1ef7e62445eece13d61030562e5a3d7a64b60606c3c12ff3be","experiment_kind":"standard","failure_count":0,"family":"spark-x2.5","generation_policy":{},"git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","model_id":"XHToken/Spark-X2.5-4B","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:8dcf43803761fd5570f637e166b655565b6c638e51a6a96848f58f6e6699a734","parameters_b":4.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.3,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:d26122294632fb1ef7e62445eece13d61030562e5a3d7a64b60606c3c12ff3be","model_key":"spark-x2.5-4b-q4km","process_cpu_percent_avg":6.338877896022156,"process_rss_bytes_avg":3725684701.8666673,"process_rss_bytes_peak":4221812736.0,"run_id":"e68754ab-25e6-4449-b06a-66dbc311d9e5","sample_count":122,"sample_interval_ms":266.60380743026815,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":8769241088.0},"run_id":"e68754ab-25e6-4449-b06a-66dbc311d9e5","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"}],"decision_summary":null,"history":[{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T15:45:57.052259+00:00","execution_signature":"sha256:execution:d26122294632fb1ef7e62445eece13d61030562e5a3d7a64b60606c3c12ff3be","failure_count":0,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:8dcf43803761fd5570f637e166b655565b6c638e51a6a96848f58f6e6699a734","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.3,"profile":"core","result_state":"CURRENT","run_id":"e68754ab-25e6-4449-b06a-66dbc311d9e5","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"model":{"artifact_format":"gguf","artifact_size_bytes":2600224352,"completed_at_utc":"2026-10-15T15:45:57.052259+00:00","deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_signature":"sha256:execution:d26122294632fb1ef7e62445eece13d61030562e5a3d7a64b60606c3c12ff3be","family":"spark-x2.5","git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","model_id":"XHToken/Spark-X2.5-4B","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:8dcf43803761fd5570f637e166b655565b6c638e51a6a96848f58f6e6699a734","parameters_b":4.0,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"e68754ab-25e6-4449-b06a-66dbc311d9e5","run_status":"COMPLETED","runtime_key":"korgis-local"},"model_signature":"sha256:model:8dcf43803761fd5570f637e166b655565b6c638e51a6a96848f58f6e6699a734","schema_version":"2"},"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866":{"current_cells":[{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T13:37:29.943647+00:00","deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.2792207792207792,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-0.8b-q4km","process_cpu_percent_avg":12.95914646037019,"process_rss_bytes_avg":4417102107.705627,"process_rss_bytes_peak":7946436608.0,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core","sample_count":462,"sample_interval_ms":199.48604018792173,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":10223632384.0},"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":532517120,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","execution_signature":"sha256:execution:c58f103ecf1090aab4510c6708cd9446fefc4137c61ae9ca33da8cc1c5d3faf8","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","parameters_b":0.8,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.1,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:c58f103ecf1090aab4510c6708cd9446fefc4137c61ae9ca33da8cc1c5d3faf8","model_key":"qwen3.5-0.8b-q4km","process_cpu_percent_avg":15.698688370479838,"process_rss_bytes_avg":1544970035.2,"process_rss_bytes_peak":2028224512.0,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":80,"sample_interval_ms":263.3856863947585,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5715886080.0},"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T13:36:13.768283+00:00","deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.2,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-0.8b-q4km","process_cpu_percent_avg":18.636242311523596,"process_rss_bytes_avg":1745210777.6,"process_rss_bytes_peak":2670067712.0,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core","sample_count":81,"sample_interval_ms":213.9264905885284,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":15259779072.0},"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T13:39:04.217888+00:00","deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.5,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-0.8b-q4km","process_cpu_percent_avg":16.11579774911694,"process_rss_bytes_avg":5182799526.2337675,"process_rss_bytes_peak":9496838144.0,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","sample_count":463,"sample_interval_ms":247.8098196266941,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":9132261376.0},"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T13:35:56.623493+00:00","deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.48333333333333334,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-0.8b-q4km","process_cpu_percent_avg":16.023245275126758,"process_rss_bytes_avg":2421507094.7555556,"process_rss_bytes_peak":3926736896.0,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core","sample_count":126,"sample_interval_ms":259.9318482166079,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":14612348928.0},"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T13:35:30.102922+00:00","deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.6833333333333333,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-0.8b-q4km","process_cpu_percent_avg":20.24050747900639,"process_rss_bytes_avg":2703435980.8,"process_rss_bytes_peak":4575264768.0,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core","sample_count":195,"sample_interval_ms":237.40003127216673,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":13519536128.0},"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"}],"decision_summary":{"artifact_format":"gguf","artifact_size_bytes":null,"current_run_ids":["2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core","2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core","2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core","2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core","b10ad834-2bdd-409b-9073-49f0233792ea"],"deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","execution_count":6,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:c58f103ecf1090aab4510c6708cd9446fefc4137c61ae9ca33da8cc1c5d3faf8","machine":"arm64","release":"25.6.0","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":3,"failure_rate":0.005905511811023622,"family":"qwen3.5","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":204788,"known_provider_cost_quality_pareto":false,"latency_mean_ms":344.54570039366,"latency_p50_ms":347.7319170051487,"latency_p95_ms":474.87177490329475,"latest_completed_at_utc":"2026-10-16T08:53:37.155498+00:00","model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","observed_case_count":508,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":true,"output_tokens_total":8771,"overall_quality_score":37.43145743145743,"parameters_b":0.8,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":508,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","machine":"arm64","release":"25.6.0","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","peak_run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","process_cpu_percent_avg":16.61227127427062,"process_rss_bytes_avg":3002504253.7158256,"process_rss_bytes_peak":9496838144.0,"resource_cell_count":6,"sample_count":1407,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5715886080.0},"result_policy":"standard","runtime_key":"korgis-local","tags":["local","open-weight","qwen","small"]},"history":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T13:37:29.943647+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.2792207792207792,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","execution_signature":"sha256:execution:c58f103ecf1090aab4510c6708cd9446fefc4137c61ae9ca33da8cc1c5d3faf8","failure_count":0,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.1,"profile":"core","result_state":"CURRENT","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T13:36:13.768283+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.2,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T13:39:04.217888+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.5,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T13:35:56.623493+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.48333333333333334,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T13:35:30.102922+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.6833333333333333,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"model":{"artifact_format":"gguf","artifact_size_bytes":532517120,"completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","execution_signature":"sha256:execution:c58f103ecf1090aab4510c6708cd9446fefc4137c61ae9ca33da8cc1c5d3faf8","family":"qwen3.5","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","parameters_b":0.8,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","run_status":"PARTIAL","runtime_key":"korgis-local"},"model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","schema_version":"2"},"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944":{"current_cells":[{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-06T09:08:01.471590+00:00","deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"xhtoken/spark-x2.5-4b","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6493506493506493,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"spark-x2.5-4b-q4km","process_cpu_percent_avg":5.594450252441077,"process_rss_bytes_avg":3797578739.069748,"process_rss_bytes_peak":6051266560.0,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core","sample_count":1308,"sample_interval_ms":310.7691591083291,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5891014656.0},"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-06T09:01:45.488881+00:00","deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"xhtoken/spark-x2.5-4b","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.45,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"spark-x2.5-4b-q4km","process_cpu_percent_avg":8.149214468504207,"process_rss_bytes_avg":3498299289.6,"process_rss_bytes_peak":3887923200.0,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core","sample_count":147,"sample_interval_ms":267.7047739123735,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":7349649408.0},"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-06T09:14:39.475593+00:00","deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"xhtoken/spark-x2.5-4b","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"spark-x2.5-4b-q4km","process_cpu_percent_avg":5.353823389553569,"process_rss_bytes_avg":4696997069.724959,"process_rss_bytes_peak":7862026240.0,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","sample_count":1416,"sample_interval_ms":302.2578173696654,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6040551424.0},"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-06T09:01:09.715786+00:00","deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"xhtoken/spark-x2.5-4b","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.95,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"spark-x2.5-4b-q4km","process_cpu_percent_avg":9.07249993238987,"process_rss_bytes_avg":3570442922.666667,"process_rss_bytes_peak":4295933952.0,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core","sample_count":244,"sample_interval_ms":280.7728714734669,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6034112512.0},"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-06T09:00:08.668486+00:00","deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"xhtoken/spark-x2.5-4b","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7666666666666667,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"spark-x2.5-4b-q4km","process_cpu_percent_avg":7.982890411325502,"process_rss_bytes_avg":3362284766.8271284,"process_rss_bytes_peak":3593027584.0,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core","sample_count":417,"sample_interval_ms":298.61884293800324,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5815320576.0},"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"}],"decision_summary":{"artifact_format":"gguf","artifact_size_bytes":2600224352,"current_run_ids":["2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core","2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core","2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core","2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core","e68754ab-25e6-4449-b06a-66dbc311d9e5"],"deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_count":6,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:d26122294632fb1ef7e62445eece13d61030562e5a3d7a64b60606c3c12ff3be","machine":"arm64","release":"25.6.0","run_id":"e68754ab-25e6-4449-b06a-66dbc311d9e5","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":1,"failure_rate":0.001968503937007874,"family":"spark-x2.5","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":234269,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1776.7870538910634,"latency_p50_ms":2137.5340625018,"latency_p95_ms":2707.341406807245,"latest_completed_at_utc":"2026-10-15T15:45:57.052259+00:00","model_id":"xhtoken/spark-x2.5-4b","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","observed_case_count":508,"observed_quality_artifact_size_pareto":true,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":9342,"overall_quality_score":67.95093795093796,"parameters_b":4,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":508,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","machine":"arm64","release":"25.6.0","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","peak_run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","process_cpu_percent_avg":7.0819593917060635,"process_rss_bytes_avg":3775214581.6258616,"process_rss_bytes_peak":7862026240.0,"resource_cell_count":6,"sample_count":3654,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5815320576.0},"result_policy":"standard","runtime_key":"korgis-local","tags":["local","open-weight","spark","reasoning","coding","agentic","q4","4b"]},"history":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-06T09:08:01.471590+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6493506493506493,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-06T09:01:45.488881+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.45,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-06T09:14:39.475593+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-06T09:01:09.715786+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.95,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-06T09:00:08.668486+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7666666666666667,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"model":{"artifact_format":"gguf","artifact_size_bytes":null,"completed_at_utc":"2026-10-06T09:14:39.475593+00:00","deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"xhtoken/spark-x2.5-4b","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","parameters_b":null,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","run_status":"COMPLETED","runtime_key":"korgis-local"},"model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","schema_version":"2"},"sha256:model:ae0099eeb8925bc828c92072ba2f486986899854d2d3de76e5689cd383c783e3":{"current_cells":[{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:718e761e4ba3410c289a5bba91a6474827caaa8da30e644b3df925ba4d16fbfc","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T16:17:35.400024+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"0065acc5513f65c3e8bbdf6899cf55e286dfe938","model_id":"qwen/qwen3.5-9b","model_key":"decisio-qwen3.5-9b-q4km","model_signature":"sha256:model:ae0099eeb8925bc828c92072ba2f486986899854d2d3de76e5689cd383c783e3","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6623376623376623,"profile":"smoke","provider_key":"decisio","quantization":"Q4_K_M","resource_summary":null,"run_id":"e9c4ec37-433f-4b06-b397-9d9b3a4f7480","runtime_key":"decisio-local","sample_count":77,"task_id":"intent-classification"}],"decision_summary":{"artifact_format":"gguf","artifact_size_bytes":null,"current_run_ids":["e9c4ec37-433f-4b06-b397-9d9b3a4f7480"],"deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_count":1,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","machine":"arm64","release":"25.6.0","run_id":"e9c4ec37-433f-4b06-b397-9d9b3a4f7480","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"smoke","failure_count":0,"failure_rate":0.0,"family":"qwen3.5","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":5000.302291662143,"latency_p50_ms":4986.015040994971,"latency_p95_ms":5194.435066799633,"latest_completed_at_utc":"2026-10-05T16:17:35.400024+00:00","model_id":"qwen/qwen3.5-9b","model_key":"decisio-qwen3.5-9b-q4km","model_signature":"sha256:model:ae0099eeb8925bc828c92072ba2f486986899854d2d3de76e5689cd383c783e3","observed_case_count":77,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":603,"overall_quality_score":66.23376623376623,"parameters_b":9,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"provider_key":"decisio","quality_coverage_capabilities":1,"quality_coverage_complete":false,"quality_coverage_datasets":1,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":null,"peak_execution_signature":null,"peak_run_id":null,"process_cpu_percent_avg":null,"process_rss_bytes_avg":null,"process_rss_bytes_peak":null,"resource_cell_count":0,"sample_count":0,"sampling_error_count":0,"scope":"unavailable","source":"unavailable","system_available_memory_bytes_min":null},"result_policy":"standard","runtime_key":"decisio-local","tags":["local","decisio","open-weight","qwen","decision-model"]},"history":[{"benchmark_signature":"sha256:benchmark:718e761e4ba3410c289a5bba91a6474827caaa8da30e644b3df925ba4d16fbfc","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T16:17:35.400024+00:00","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","failure_count":0,"model_key":"decisio-qwen3.5-9b-q4km","model_signature":"sha256:model:ae0099eeb8925bc828c92072ba2f486986899854d2d3de76e5689cd383c783e3","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6623376623376623,"profile":"smoke","result_state":"CURRENT","run_id":"e9c4ec37-433f-4b06-b397-9d9b3a4f7480","sample_count":77,"status":"COMPLETED","task_id":"intent-classification"}],"model":{"artifact_format":"gguf","artifact_size_bytes":null,"completed_at_utc":"2026-10-05T16:17:35.400024+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","family":null,"git_commit":"0065acc5513f65c3e8bbdf6899cf55e286dfe938","model_id":"qwen/qwen3.5-9b","model_key":"decisio-qwen3.5-9b-q4km","model_signature":"sha256:model:ae0099eeb8925bc828c92072ba2f486986899854d2d3de76e5689cd383c783e3","parameters_b":null,"provider_key":"decisio","quantization":"Q4_K_M","run_id":"e9c4ec37-433f-4b06-b397-9d9b3a4f7480","run_status":"COMPLETED","runtime_key":"decisio-local"},"model_signature":"sha256:model:ae0099eeb8925bc828c92072ba2f486986899854d2d3de76e5689cd383c783e3","schema_version":"2"},"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce":{"current_cells":[{"artifact_format":"gguf","artifact_size_bytes":10947100928,"benchmark_signature":"sha256:consolidated:42ab94943101512840504e431240a2dc1e9c2c51d6e055498c7b6b44c803f515","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-22T13:12:52.242438+00:00","deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","experiment_kind":"consolidated","failure_count":0,"family":"lfm2.5","generation_policy":{"generation_configs":[{"extra":{},"max_output_tokens":2048,"seed":42,"stop":[],"temperature":0.0},{"extra":{},"max_output_tokens":4096,"seed":42,"stop":[],"temperature":0.0}],"mixed_generation":true},"git_commit":null,"model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.7012987012987013,"profile":"core","provider_key":"korgis","quantization":"UD-Q3_K_XL","resource_summary":null,"run_id":"lfm25-consolidated-latest","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":10947100928,"benchmark_signature":"sha256:consolidated:20f65d6394555078bc9dce0c1393e10d963aecb84ce9344ece12046a78d42de3","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-22T13:12:52.242438+00:00","deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","experiment_kind":"consolidated","failure_count":0,"family":"lfm2.5","generation_policy":{"generation_configs":[{"extra":{},"max_output_tokens":2048,"seed":42,"stop":[],"temperature":0.0},{"extra":{},"max_output_tokens":4096,"seed":42,"stop":[],"temperature":0.0}],"mixed_generation":true},"git_commit":null,"model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.95,"profile":"core","provider_key":"korgis","quantization":"UD-Q3_K_XL","resource_summary":null,"run_id":"lfm25-consolidated-latest","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":10947100928,"benchmark_signature":"sha256:benchmark:600e8b1d94c1e2bc55bd8802918ce0c6f18dece6d8a9c02247a530fc7b015409","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-22T14:07:00.510877+00:00","deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_signature":"sha256:execution:503d302009e21bb18588b87ac3897f9112994a479905bbca872384efa4360298","experiment_kind":"standard","failure_count":0,"family":"lfm2.5","generation_policy":{"max_output_tokens":4096},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.775,"profile":"core","provider_key":"korgis","quantization":"UD-Q3_K_XL","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:503d302009e21bb18588b87ac3897f9112994a479905bbca872384efa4360298","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","process_cpu_percent_avg":10.29314878109045,"process_rss_bytes_avg":1417688467.7424607,"process_rss_bytes_peak":1463648256.0,"run_id":"lfm25-math-v2-core","sample_count":1316,"sample_interval_ms":313.23082923487743,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4090920960.0},"run_id":"lfm25-math-v2-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":10947100928,"benchmark_signature":"sha256:consolidated:9468d75aebe9e805cefc3643f61a9e5fe6127c7a4e3a987608b30841b9bc9fa0","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-22T13:12:52.242438+00:00","deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","experiment_kind":"consolidated","failure_count":0,"family":"lfm2.5","generation_policy":{"generation_configs":[{"extra":{},"max_output_tokens":2048,"seed":42,"stop":[],"temperature":0.0},{"extra":{},"max_output_tokens":4096,"seed":42,"stop":[],"temperature":0.0}],"mixed_generation":true},"git_commit":null,"model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.9545454545454546,"profile":"core","provider_key":"korgis","quantization":"UD-Q3_K_XL","resource_summary":null,"run_id":"lfm25-consolidated-latest","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":10947100928,"benchmark_signature":"sha256:consolidated:b832fe612fcfe138845fe193a557c686f68ecb2f59d1b3f4784e653e9ec82e5c","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-22T13:12:52.242438+00:00","deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","experiment_kind":"consolidated","failure_count":0,"family":"lfm2.5","generation_policy":{"generation_configs":[{"extra":{},"max_output_tokens":2048,"seed":42,"stop":[],"temperature":0.0},{"extra":{},"max_output_tokens":4096,"seed":42,"stop":[],"temperature":0.0}],"mixed_generation":true},"git_commit":null,"model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"profile":"core","provider_key":"korgis","quantization":"UD-Q3_K_XL","resource_summary":null,"run_id":"lfm25-consolidated-latest","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":10947100928,"benchmark_signature":"sha256:consolidated:fd230e4e9f030196ab13d72a0beb260f92fbe3bfb1cfa331ed8e0678b55123f9","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-22T13:12:52.242438+00:00","deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","experiment_kind":"consolidated","failure_count":0,"family":"lfm2.5","generation_policy":{"generation_configs":[{"extra":{},"max_output_tokens":2048,"seed":42,"stop":[],"temperature":0.0},{"extra":{},"max_output_tokens":4096,"seed":42,"stop":[],"temperature":0.0}],"mixed_generation":true},"git_commit":null,"model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7666666666666667,"profile":"core","provider_key":"korgis","quantization":"UD-Q3_K_XL","resource_summary":null,"run_id":"lfm25-consolidated-latest","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"}],"decision_summary":{"artifact_format":"gguf","artifact_size_bytes":10947100928,"current_run_ids":["lfm25-consolidated-latest","lfm25-math-v2-core"],"deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_count":2,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:503d302009e21bb18588b87ac3897f9112994a479905bbca872384efa4360298","machine":"arm64","release":"25.6.0","run_id":"lfm25-math-v2-core","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":15,"failure_rate":0.02952755905511811,"family":"lfm2.5","generation_parameters":{"extra":{},"max_output_tokens":2048,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":223016,"known_provider_cost_quality_pareto":false,"latency_mean_ms":16482.3864442058,"latency_p50_ms":9803.956041461788,"latency_p95_ms":60905.82317115038,"latest_completed_at_utc":"2026-10-22T14:07:00.510877+00:00","model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","observed_case_count":508,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":true,"output_tokens_total":373453,"overall_quality_score":85.23629148629148,"parameters_b":24.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":508,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"UD-Q3_K_XL","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:503d302009e21bb18588b87ac3897f9112994a479905bbca872384efa4360298","machine":"arm64","release":"25.6.0","run_id":"lfm25-math-v2-core","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:503d302009e21bb18588b87ac3897f9112994a479905bbca872384efa4360298","peak_run_id":"lfm25-math-v2-core","process_cpu_percent_avg":10.29314878109045,"process_rss_bytes_avg":1417688467.7424607,"process_rss_bytes_peak":1463648256.0,"resource_cell_count":1,"sample_count":1316,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4090920960.0},"result_policy":"consolidated-latest-attempt","runtime_key":"korgis-local","tags":["local","open-weight","moe"]},"history":[{"benchmark_signature":"sha256:benchmark:d6ab2ffb1f34c76d5dddef81996eb87dc7f2d20e82b7d6e34eff857d6c17bb72","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-22T09:44:42.382206+00:00","execution_signature":"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6493506493506493,"profile":"core","result_state":"CURRENT","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"},{"benchmark_signature":"sha256:benchmark:3e83e05786bfa67a0e16cb28b57ec52fff1967e5e59a7bb032185b9df349aa2d","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-22T12:49:48.793583+00:00","execution_signature":"sha256:execution:641822ff7e4fc678a816a765bd43258a89b949aad2eec0a36c6fb22e990053ee","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.2857142857142857,"profile":"core","result_state":"HISTORICAL","run_id":"lfm25-recovery-4096","sample_count":28,"status":"COMPLETED","task_id":"intent-classification"},{"benchmark_signature":"sha256:consolidated:42ab94943101512840504e431240a2dc1e9c2c51d6e055498c7b6b44c803f515","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-22T13:12:52.242438+00:00","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.7012987012987013,"profile":"core","result_state":"HISTORICAL","run_id":"lfm25-consolidated-latest","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"},{"benchmark_signature":"sha256:benchmark:a5db4a5d2c68fabba1fda0494f04219381d92bf682959ba51aed96f18f3605d2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-22T08:51:38.368018+00:00","execution_signature":"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.95,"profile":"core","result_state":"CURRENT","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:consolidated:20f65d6394555078bc9dce0c1393e10d963aecb84ce9344ece12046a78d42de3","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-22T13:12:52.242438+00:00","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.95,"profile":"core","result_state":"HISTORICAL","run_id":"lfm25-consolidated-latest","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:600e8b1d94c1e2bc55bd8802918ce0c6f18dece6d8a9c02247a530fc7b015409","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-22T14:07:00.510877+00:00","execution_signature":"sha256:execution:503d302009e21bb18588b87ac3897f9112994a479905bbca872384efa4360298","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.775,"profile":"core","result_state":"CURRENT","run_id":"lfm25-math-v2-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:230c029eec827b8bb748e50afe4cc525fae176746626e1dc4cb43e9743332e65","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-22T10:24:06.430109+00:00","execution_signature":"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.974025974025974,"profile":"core","result_state":"CURRENT","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"},{"benchmark_signature":"sha256:benchmark:b7c8f155e579d1d1d981fb2d54369194d1a8415fe5833ed759f937156c77e621","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-22T12:49:48.793583+00:00","execution_signature":"sha256:execution:641822ff7e4fc678a816a765bd43258a89b949aad2eec0a36c6fb22e990053ee","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.8,"profile":"core","result_state":"HISTORICAL","run_id":"lfm25-recovery-4096","sample_count":20,"status":"COMPLETED","task_id":"calibrated-intent-classification"},{"benchmark_signature":"sha256:consolidated:9468d75aebe9e805cefc3643f61a9e5fe6127c7a4e3a987608b30841b9bc9fa0","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-22T13:12:52.242438+00:00","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.9545454545454546,"profile":"core","result_state":"HISTORICAL","run_id":"lfm25-consolidated-latest","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"},{"benchmark_signature":"sha256:benchmark:5ef83b758c4b66715eca88f843e6e438a1f602661374de3eac156fc10ef9f203","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-22T08:45:22.525820+00:00","execution_signature":"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"profile":"core","result_state":"CURRENT","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"},{"benchmark_signature":"sha256:consolidated:b832fe612fcfe138845fe193a557c686f68ecb2f59d1b3f4784e653e9ec82e5c","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-22T13:12:52.242438+00:00","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"profile":"core","result_state":"HISTORICAL","run_id":"lfm25-consolidated-latest","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"},{"benchmark_signature":"sha256:benchmark:b8edb9ae85ac7482f2930cdbf8920bc573cea8dded0851e3351b0e05a8900d88","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-22T08:16:14.436917+00:00","execution_signature":"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7666666666666667,"profile":"core","result_state":"CURRENT","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"},{"benchmark_signature":"sha256:benchmark:c26dc80a4a3d7172c20aaf4a0d946e31de5730fcbc2b0e3ef1656e2f69c25727","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-22T12:49:48.793583+00:00","execution_signature":"sha256:execution:641822ff7e4fc678a816a765bd43258a89b949aad2eec0a36c6fb22e990053ee","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.0,"profile":"core","result_state":"HISTORICAL","run_id":"lfm25-recovery-4096","sample_count":1,"status":"COMPLETED","task_id":"structured-output"},{"benchmark_signature":"sha256:consolidated:fd230e4e9f030196ab13d72a0beb260f92fbe3bfb1cfa331ed8e0678b55123f9","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-22T13:12:52.242438+00:00","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7666666666666667,"profile":"core","result_state":"HISTORICAL","run_id":"lfm25-consolidated-latest","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"model":{"artifact_format":"gguf","artifact_size_bytes":10947100928,"completed_at_utc":"2026-10-22T14:07:00.510877+00:00","deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_signature":"sha256:execution:503d302009e21bb18588b87ac3897f9112994a479905bbca872384efa4360298","family":"lfm2.5","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"provider_key":"korgis","quantization":"UD-Q3_K_XL","run_id":"lfm25-math-v2-core","run_status":"COMPLETED","runtime_key":"korgis-local"},"model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","schema_version":"2"},"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8":{"current_cells":[{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-06T12:22:05.300981+00:00","deployment":"api","effective_model_id":"gpt-5.6-luna","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"11b00f55c776ec89ddee7cd2dd1898797dde4055","model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.8246753246753247,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-intent-classification-core","runtime_key":"openai-api","sample_count":154,"task_id":"intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T15:34:31.093143+00:00","deployment":"api","effective_model_id":"gpt-5.6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","experiment_kind":"standard","failure_count":0,"family":"gpt-5.6","generation_policy":{},"git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.675,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","runtime_key":"openai-api","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-06T12:17:36.689339+00:00","deployment":"api","effective_model_id":"gpt-5.6-luna","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.45,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core","runtime_key":"openai-api","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-06T12:26:04.935243+00:00","deployment":"api","effective_model_id":"gpt-5.6-luna","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"11b00f55c776ec89ddee7cd2dd1898797dde4055","model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":1.0,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core","runtime_key":"openai-api","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-06T12:16:26.396465+00:00","deployment":"api","effective_model_id":"gpt-5.6-luna","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core","runtime_key":"openai-api","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-06T12:15:04.761551+00:00","deployment":"api","effective_model_id":"gpt-5.6-luna","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.9166666666666666,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-structured-output-core","runtime_key":"openai-api","sample_count":60,"task_id":"structured-output"}],"decision_summary":{"artifact_format":null,"artifact_size_bytes":null,"current_run_ids":["2026-10-core-v1-gpt-5.6-luna-intent-classification-core","2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core","2026-10-core-v1-gpt-5.6-luna-oos-calibration-core","2026-10-core-v1-gpt-5.6-luna-qa-abstention-core","2026-10-core-v1-gpt-5.6-luna-structured-output-core","f56a81bc-d911-4b76-9b68-361b14e92a2c"],"deployment":"api","effective_model_id":"gpt-5.6-luna","execution_count":6,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","machine":"arm64","release":"25.6.0","run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":8,"failure_rate":0.015748031496062992,"family":"gpt-5.6","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":334792,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1649.373420041974,"latency_p50_ms":1445.532062505663,"latency_p95_ms":2903.054808448358,"latest_completed_at_utc":"2026-10-15T15:34:31.093143+00:00","model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","observed_case_count":508,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":18155,"overall_quality_score":78.88347763347763,"parameters_b":null,"provider_cost_coverage_rate":0.984251968503937,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0887444,"provider_cost_per_1k_cases_usd":0.1774888,"provider_cost_per_case_usd":0.0001774888,"provider_cost_priced_cases":500,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"openai","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":null,"resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":null,"peak_execution_signature":null,"peak_run_id":null,"process_cpu_percent_avg":null,"process_rss_bytes_avg":null,"process_rss_bytes_peak":null,"resource_cell_count":0,"sample_count":0,"sampling_error_count":0,"scope":"unavailable","source":"unavailable","system_available_memory_bytes_min":null},"result_policy":"standard","runtime_key":"openai-api","tags":["api","closed","openai"]},"history":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-06T12:22:05.300981+00:00","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","failure_count":0,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.8246753246753247,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-gpt-5.6-luna-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T15:34:31.093143+00:00","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":0,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.675,"profile":"core","result_state":"CURRENT","run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-06T12:17:36.689339+00:00","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","failure_count":0,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.45,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-06T12:26:04.935243+00:00","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","failure_count":0,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":1.0,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-06T12:16:26.396465+00:00","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","failure_count":0,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-06T12:15:04.761551+00:00","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","failure_count":0,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.9166666666666666,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-gpt-5.6-luna-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"model":{"artifact_format":null,"artifact_size_bytes":null,"completed_at_utc":"2026-10-15T15:34:31.093143+00:00","deployment":"api","effective_model_id":"gpt-5.6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","family":"gpt-5.6","git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","parameters_b":null,"provider_key":"openai","quantization":null,"run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","run_status":"COMPLETED","runtime_key":"openai-api"},"model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","schema_version":"2"},"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e":{"current_cells":[{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T13:39:20.878424+00:00","deployment":"api","effective_model_id":"gemini-3.8-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.8","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.8-flash","model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.0,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core","runtime_key":"gemini-api","sample_count":154,"task_id":"intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T13:38:58.640898+00:00","deployment":"api","effective_model_id":"gemini-3.8-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.8","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.8-flash","model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.0,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core","runtime_key":"gemini-api","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T13:39:43.452981+00:00","deployment":"api","effective_model_id":"gemini-3.8-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.8","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.8-flash","model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.5,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core","runtime_key":"gemini-api","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T13:38:52.091017+00:00","deployment":"api","effective_model_id":"gemini-3.8-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.8","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.8-flash","model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.0,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core","runtime_key":"gemini-api","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T13:38:42.203597+00:00","deployment":"api","effective_model_id":"gemini-3.8-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.8","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.8-flash","model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.0,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core","runtime_key":"gemini-api","sample_count":60,"task_id":"structured-output"}],"decision_summary":{"artifact_format":null,"artifact_size_bytes":null,"current_run_ids":["gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core","gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core","gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core","gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core","gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core"],"deployment":"api","effective_model_id":"gemini-3.8-flash","execution_count":5,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","machine":"arm64","release":"25.6.0","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":468,"failure_rate":1.0,"family":"gemini-3.8","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":754.4821022909471,"latency_p50_ms":129.50879096752033,"latency_p95_ms":208.09100856131408,"latest_completed_at_utc":"2026-10-13T13:39:43.452981+00:00","model_id":"gemini-3.8-flash","model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","observed_case_count":468,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":true,"observed_quality_parameters_pareto":false,"output_tokens_total":null,"overall_quality_score":10.0,"parameters_b":null,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.8-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":468,"provider_cost_total_usd":null,"provider_key":"gemini","quality_coverage_capabilities":5,"quality_coverage_complete":false,"quality_coverage_datasets":5,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":null,"resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":null,"peak_execution_signature":null,"peak_run_id":null,"process_cpu_percent_avg":null,"process_rss_bytes_avg":null,"process_rss_bytes_peak":null,"resource_cell_count":0,"sample_count":0,"sampling_error_count":0,"scope":"unavailable","source":"unavailable","system_available_memory_bytes_min":null},"result_policy":"standard","runtime_key":"gemini-api","tags":["api","closed","google","gemini","flash"]},"history":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T13:39:20.878424+00:00","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":0,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.0,"profile":"core","result_state":"CURRENT","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T13:38:58.640898+00:00","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":0,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.0,"profile":"core","result_state":"CURRENT","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T13:39:43.452981+00:00","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":0,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.5,"profile":"core","result_state":"CURRENT","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T13:38:52.091017+00:00","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":0,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.0,"profile":"core","result_state":"CURRENT","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T13:38:42.203597+00:00","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":0,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.0,"profile":"core","result_state":"CURRENT","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"model":{"artifact_format":null,"artifact_size_bytes":null,"completed_at_utc":"2026-10-13T13:39:43.452981+00:00","deployment":"api","effective_model_id":"gemini-3.8-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","family":"gemini-3.8","git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.8-flash","model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","parameters_b":null,"provider_key":"gemini","quantization":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core","run_status":"COMPLETED","runtime_key":"gemini-api"},"model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","schema_version":"2"},"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42":{"current_cells":[{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T12:46:44.225239+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6948051948051948,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-9b-q4km","process_cpu_percent_avg":3.9138417761512048,"process_rss_bytes_avg":6965862817.821014,"process_rss_bytes_peak":10559422464.0,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core","sample_count":1620,"sample_interval_ms":303.4210088575144,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5489934336.0},"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":5627044256,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:9d728f687c9bb0f47a2c5a8c0e61b6da5a7e2833ad1afcffb18c7b4eb2767886","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","parameters_b":9.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.5,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:9d728f687c9bb0f47a2c5a8c0e61b6da5a7e2833ad1afcffb18c7b4eb2767886","model_key":"qwen3.5-9b-q4km","process_cpu_percent_avg":7.04412305796482,"process_rss_bytes_avg":8143566028.8,"process_rss_bytes_peak":9486303232.0,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":176,"sample_interval_ms":291.7408815034529,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":9481781248.0},"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T12:39:03.216607+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.675,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-9b-q4km","process_cpu_percent_avg":6.526783901766963,"process_rss_bytes_avg":8611465482.24,"process_rss_bytes_peak":11088855040.0,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core","sample_count":175,"sample_interval_ms":275.92921112542774,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":11636277248.0},"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T13:01:17.635848+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.974025974025974,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-9b-q4km","process_cpu_percent_avg":2.683081412939199,"process_rss_bytes_avg":6844874999.319381,"process_rss_bytes_peak":10287939584.0,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core","sample_count":2848,"sample_interval_ms":314.6654996565356,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5142609920.0},"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T12:38:16.531029+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.9333333333333333,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-9b-q4km","process_cpu_percent_avg":6.3201481230234045,"process_rss_bytes_avg":8711409521.940317,"process_rss_bytes_peak":11538563072.0,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","sample_count":345,"sample_interval_ms":281.76636477524414,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":7018201088.0},"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T12:36:44.930375+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7666666666666667,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-9b-q4km","process_cpu_percent_avg":5.8239325570441824,"process_rss_bytes_avg":6475378858.969259,"process_rss_bytes_peak":7243612160.0,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core","sample_count":617,"sample_interval_ms":298.72389312365516,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5452595200.0},"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"}],"decision_summary":{"artifact_format":"gguf","artifact_size_bytes":null,"current_run_ids":["2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core","2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core","2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core","2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core","b10ad834-2bdd-409b-9073-49f0233792ea"],"deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_count":6,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:9d728f687c9bb0f47a2c5a8c0e61b6da5a7e2833ad1afcffb18c7b4eb2767886","machine":"arm64","release":"25.6.0","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":2,"failure_rate":0.003937007874015748,"family":"qwen3.5","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":204788,"known_provider_cost_quality_pareto":false,"latency_mean_ms":3119.1373101190284,"latency_p50_ms":2776.106687495485,"latency_p95_ms":7684.784721046889,"latest_completed_at_utc":"2026-10-16T08:53:37.155498+00:00","model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","observed_case_count":508,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":true,"output_tokens_total":8355,"overall_quality_score":75.73051948051948,"parameters_b":9,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":508,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","machine":"arm64","release":"25.6.0","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","peak_run_id":"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","process_cpu_percent_avg":5.385318471481629,"process_rss_bytes_avg":7625426284.848329,"process_rss_bytes_peak":11538563072.0,"resource_cell_count":6,"sample_count":5781,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5142609920.0},"result_policy":"standard","runtime_key":"korgis-local","tags":["local","open-weight","qwen"]},"history":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T12:46:44.225239+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6948051948051948,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","execution_signature":"sha256:execution:9d728f687c9bb0f47a2c5a8c0e61b6da5a7e2833ad1afcffb18c7b4eb2767886","failure_count":0,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.5,"profile":"core","result_state":"CURRENT","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T12:39:03.216607+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.675,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T13:01:17.635848+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.974025974025974,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T12:38:16.531029+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.9333333333333333,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T12:36:44.930375+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7666666666666667,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"model":{"artifact_format":"gguf","artifact_size_bytes":5627044256,"completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:9d728f687c9bb0f47a2c5a8c0e61b6da5a7e2833ad1afcffb18c7b4eb2767886","family":"qwen3.5","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","parameters_b":9.0,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","run_status":"PARTIAL","runtime_key":"korgis-local"},"model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","schema_version":"2"},"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3":{"current_cells":[{"artifact_format":"gguf","artifact_size_bytes":4482403488,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T11:37:13.963536+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q8","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","parameters_b":4.0,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6493506493506493,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","model_key":"qwen3.5-4b-q8","process_cpu_percent_avg":5.113588871923991,"process_rss_bytes_avg":7039335235.285094,"process_rss_bytes_peak":10390700032.0,"run_id":"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core","sample_count":1041,"sample_interval_ms":311.43875954839615,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5581979648.0},"run_id":"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":4482403488,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q8","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","parameters_b":4.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.475,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","model_key":"qwen3.5-4b-q8","process_cpu_percent_avg":10.86561479591514,"process_rss_bytes_avg":5559348804.266666,"process_rss_bytes_peak":6319587328.0,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":146,"sample_interval_ms":295.3429861280407,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4841750528.0},"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":4482403488,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T11:32:22.918386+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q8","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","parameters_b":4.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.5,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","model_key":"qwen3.5-4b-q8","process_cpu_percent_avg":6.982037738154942,"process_rss_bytes_avg":7516327697.066668,"process_rss_bytes_peak":9962160128.0,"run_id":"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core","sample_count":132,"sample_interval_ms":291.2576532937237,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":11326062592.0},"run_id":"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":4482403488,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T11:43:07.482305+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q8","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","parameters_b":4.0,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.9805194805194806,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","model_key":"qwen3.5-4b-q8","process_cpu_percent_avg":5.610485336009209,"process_rss_bytes_avg":7096272805.442177,"process_rss_bytes_peak":11637391360.0,"run_id":"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","sample_count":1211,"sample_interval_ms":319.76690310093494,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5589811200.0},"run_id":"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":4482403488,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T11:31:50.661853+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q8","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","parameters_b":4.0,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","model_key":"qwen3.5-4b-q8","process_cpu_percent_avg":6.023120834777095,"process_rss_bytes_avg":7696510827.330371,"process_rss_bytes_peak":10978181120.0,"run_id":"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core","sample_count":278,"sample_interval_ms":297.275809231578,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":7347224576.0},"run_id":"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":4482403488,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T11:30:38.611895+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q8","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","parameters_b":4.0,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7333333333333333,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","model_key":"qwen3.5-4b-q8","process_cpu_percent_avg":5.698576220486698,"process_rss_bytes_avg":6661104577.102827,"process_rss_bytes_peak":9700769792.0,"run_id":"qwen-variants-core-qwen3.5-4b-q8-structured-output-core","sample_count":477,"sample_interval_ms":317.01940462127334,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5716574208.0},"run_id":"qwen-variants-core-qwen3.5-4b-q8-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"}],"decision_summary":{"artifact_format":"gguf","artifact_size_bytes":4482403488,"current_run_ids":["b10ad834-2bdd-409b-9073-49f0233792ea","qwen-variants-core-qwen3.5-4b-q8-intent-classification-core","qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core","qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core","qwen-variants-core-qwen3.5-4b-q8-structured-output-core"],"deployment":"local","effective_model_id":"qwen3.5-4b-q8","execution_count":6,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","machine":"arm64","release":"25.6.0","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":2,"failure_rate":0.003937007874015748,"family":"qwen3.5","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":204788,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1605.0692064233635,"latency_p50_ms":1699.8242709960323,"latency_p95_ms":2421.302295583882,"latest_completed_at_utc":"2026-10-16T08:53:37.155498+00:00","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","observed_case_count":508,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":8488,"overall_quality_score":70.08116883116882,"parameters_b":4.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":508,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","machine":"arm64","release":"25.6.0","run_id":"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","peak_run_id":"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","process_cpu_percent_avg":6.715570632877846,"process_rss_bytes_avg":6928149991.082301,"process_rss_bytes_peak":11637391360.0,"resource_cell_count":6,"sample_count":3285,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4841750528.0},"result_policy":"standard","runtime_key":"korgis-local","tags":["local","open-weight","qwen","4b","q8"]},"history":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T11:37:13.963536+00:00","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","failure_count":0,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6493506493506493,"profile":"core","result_state":"CURRENT","run_id":"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","failure_count":0,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.475,"profile":"core","result_state":"CURRENT","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T11:32:22.918386+00:00","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","failure_count":0,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.5,"profile":"core","result_state":"CURRENT","run_id":"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T11:43:07.482305+00:00","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","failure_count":0,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.9805194805194806,"profile":"core","result_state":"CURRENT","run_id":"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T11:31:50.661853+00:00","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","failure_count":0,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"profile":"core","result_state":"CURRENT","run_id":"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T11:30:38.611895+00:00","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","failure_count":0,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7333333333333333,"profile":"core","result_state":"CURRENT","run_id":"qwen-variants-core-qwen3.5-4b-q8-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"model":{"artifact_format":"gguf","artifact_size_bytes":4482403488,"completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q8","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","family":"qwen3.5","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","parameters_b":4.0,"provider_key":"korgis","quantization":"Q8_0","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","run_status":"PARTIAL","runtime_key":"korgis-local"},"model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","schema_version":"2"},"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694":{"current_cells":[{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T11:25:37.855179+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.577922077922078,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-2b-q4km","process_cpu_percent_avg":9.421328984702535,"process_rss_bytes_avg":2155211762.7012987,"process_rss_bytes_peak":3131637760.0,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core","sample_count":616,"sample_interval_ms":260.0283379352414,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6455427072.0},"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":1280835840,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T15:44:53.050810+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:8190c8f3bb1f7b51ef597577c913ceae05f8cb7998a7d5221e848f67071166c9","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","parameters_b":2.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.25,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:8190c8f3bb1f7b51ef597577c913ceae05f8cb7998a7d5221e848f67071166c9","model_key":"qwen3.5-2b-q4km","process_cpu_percent_avg":8.953256029208966,"process_rss_bytes_avg":2160430830.9333334,"process_rss_bytes_peak":2483290112.0,"run_id":"9fb9f932-1f29-4646-b337-f33476d841ba","sample_count":94,"sample_interval_ms":267.4467081844341,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5225758720.0},"run_id":"9fb9f932-1f29-4646-b337-f33476d841ba","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T11:23:17.557837+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.325,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-2b-q4km","process_cpu_percent_avg":12.866597281256125,"process_rss_bytes_avg":2496888695.466666,"process_rss_bytes_peak":3418226688.0,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core","sample_count":104,"sample_interval_ms":243.8169525759804,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":8507686912.0},"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T11:28:16.098461+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.5,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-2b-q4km","process_cpu_percent_avg":8.346543248013683,"process_rss_bytes_avg":3164761335.3558445,"process_rss_bytes_peak":6107152384.0,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","sample_count":621,"sample_interval_ms":291.97748509630554,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6536364032.0},"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T11:22:51.387753+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.7833333333333333,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-2b-q4km","process_cpu_percent_avg":14.926939453226597,"process_rss_bytes_avg":2006719510.7555552,"process_rss_bytes_peak":2334687232.0,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core","sample_count":182,"sample_interval_ms":249.7748705191043,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6527320064.0},"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T10:55:49.921242+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7333333333333333,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-2b-q4km","process_cpu_percent_avg":15.462938913292911,"process_rss_bytes_avg":2241664864.711111,"process_rss_bytes_peak":2811346944.0,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core","sample_count":253,"sample_interval_ms":275.0343119529134,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6283788288.0},"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"}],"decision_summary":{"artifact_format":"gguf","artifact_size_bytes":null,"current_run_ids":["2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core","2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core","2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core","2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core","9fb9f932-1f29-4646-b337-f33476d841ba"],"deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_count":6,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:8190c8f3bb1f7b51ef597577c913ceae05f8cb7998a7d5221e848f67071166c9","machine":"arm64","release":"25.6.0","run_id":"9fb9f932-1f29-4646-b337-f33476d841ba","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":4,"failure_rate":0.007874015748031496,"family":"qwen3.5","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":204788,"known_provider_cost_quality_pareto":false,"latency_mean_ms":646.8228120271721,"latency_p50_ms":700.4097294993699,"latency_p95_ms":879.9385391488613,"latest_completed_at_utc":"2026-10-15T15:44:53.050810+00:00","model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","observed_case_count":508,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":8820,"overall_quality_score":52.826479076479075,"parameters_b":2,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":508,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","machine":"arm64","release":"25.6.0","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","peak_run_id":"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","process_cpu_percent_avg":11.662933984950136,"process_rss_bytes_avg":2370946166.6539683,"process_rss_bytes_peak":6107152384.0,"resource_cell_count":6,"sample_count":1870,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5225758720.0},"result_policy":"standard","runtime_key":"korgis-local","tags":["local","open-weight","qwen","small"]},"history":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T11:25:37.855179+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.577922077922078,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T15:44:53.050810+00:00","execution_signature":"sha256:execution:8190c8f3bb1f7b51ef597577c913ceae05f8cb7998a7d5221e848f67071166c9","failure_count":0,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.25,"profile":"core","result_state":"CURRENT","run_id":"9fb9f932-1f29-4646-b337-f33476d841ba","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T11:23:17.557837+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.325,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T11:28:16.098461+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.5,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T11:22:51.387753+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.7833333333333333,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T10:55:49.921242+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7333333333333333,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"model":{"artifact_format":"gguf","artifact_size_bytes":1280835840,"completed_at_utc":"2026-10-15T15:44:53.050810+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:8190c8f3bb1f7b51ef597577c913ceae05f8cb7998a7d5221e848f67071166c9","family":"qwen3.5","git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","parameters_b":2.0,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"9fb9f932-1f29-4646-b337-f33476d841ba","run_status":"COMPLETED","runtime_key":"korgis-local"},"model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","schema_version":"2"},"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d":{"current_cells":[{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-06T10:03:19.101787+00:00","deployment":"local","effective_model_id":"bonsai-27b-q10","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.5974025974025974,"profile":"core","provider_key":"korgis","quantization":"Q1_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"bonsai-27b-q10","process_cpu_percent_avg":4.141997751956015,"process_rss_bytes_avg":5899303355.703257,"process_rss_bytes_peak":10491068416.0,"run_id":"2026-10-core-v1-bonsai-27b-q10-intent-classification-core","sample_count":4416,"sample_interval_ms":317.0853588508965,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4949098496.0},"run_id":"2026-10-core-v1-bonsai-27b-q10-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":3803452480,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"bonsai-27b-q10","execution_signature":"sha256:execution:80688f79e2c27746cabb9c8e268a98d63ee3d732cb5c73d70fc581a1958017fc","experiment_kind":"standard","failure_count":0,"family":"bonsai","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","parameters_b":27.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.475,"profile":"core","provider_key":"korgis","quantization":"Q1_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:80688f79e2c27746cabb9c8e268a98d63ee3d732cb5c73d70fc581a1958017fc","model_key":"bonsai-27b-q10","process_cpu_percent_avg":10.004219049002119,"process_rss_bytes_avg":7178716753.609695,"process_rss_bytes_peak":9109520384.0,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":342,"sample_interval_ms":299.5298028563516,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5797576704.0},"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-06T09:40:29.476460+00:00","deployment":"local","effective_model_id":"bonsai-27b-q10","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.575,"profile":"core","provider_key":"korgis","quantization":"Q1_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"bonsai-27b-q10","process_cpu_percent_avg":10.398893667073832,"process_rss_bytes_avg":7636308908.292063,"process_rss_bytes_peak":11685068800.0,"run_id":"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","sample_count":340,"sample_interval_ms":307.1656581635226,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5910167552.0},"run_id":"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-06T10:26:04.512560+00:00","deployment":"local","effective_model_id":"bonsai-27b-q10","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.948051948051948,"profile":"core","provider_key":"korgis","quantization":"Q1_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"bonsai-27b-q10","process_cpu_percent_avg":3.871693631322875,"process_rss_bytes_avg":5867857169.246841,"process_rss_bytes_peak":9145516032.0,"run_id":"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core","sample_count":4524,"sample_interval_ms":309.6376288065501,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5604933632.0},"run_id":"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-06T09:38:49.212082+00:00","deployment":"local","effective_model_id":"bonsai-27b-q10","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"profile":"core","provider_key":"korgis","quantization":"Q1_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"bonsai-27b-q10","process_cpu_percent_avg":9.839476147142088,"process_rss_bytes_avg":6890131131.833357,"process_rss_bytes_peak":10222600192.0,"run_id":"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core","sample_count":690,"sample_interval_ms":311.07264410993747,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5791531008.0},"run_id":"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-06T09:35:22.771531+00:00","deployment":"local","effective_model_id":"bonsai-27b-q10","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7833333333333333,"profile":"core","provider_key":"korgis","quantization":"Q1_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"bonsai-27b-q10","process_cpu_percent_avg":9.52518999663954,"process_rss_bytes_avg":6054163284.918202,"process_rss_bytes_peak":8022573056.0,"run_id":"2026-10-core-v1-bonsai-27b-q10-structured-output-core","sample_count":990,"sample_interval_ms":317.9005326878062,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5717721088.0},"run_id":"2026-10-core-v1-bonsai-27b-q10-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"}],"decision_summary":{"artifact_format":"gguf","artifact_size_bytes":null,"current_run_ids":["2026-10-core-v1-bonsai-27b-q10-intent-classification-core","2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","2026-10-core-v1-bonsai-27b-q10-oos-calibration-core","2026-10-core-v1-bonsai-27b-q10-qa-abstention-core","2026-10-core-v1-bonsai-27b-q10-structured-output-core","b10ad834-2bdd-409b-9073-49f0233792ea"],"deployment":"local","effective_model_id":"bonsai-27b-q10","execution_count":6,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:80688f79e2c27746cabb9c8e268a98d63ee3d732cb5c73d70fc581a1958017fc","machine":"arm64","release":"25.6.0","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":6,"failure_rate":0.011811023622047244,"family":"bonsai","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":204788,"known_provider_cost_quality_pareto":false,"latency_mean_ms":6572.755784444501,"latency_p50_ms":8354.460334005125,"latency_p95_ms":9424.563083646353,"latest_completed_at_utc":"2026-10-16T08:53:37.155498+00:00","model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","observed_case_count":508,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":9910,"overall_quality_score":70.75757575757575,"parameters_b":27,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":508,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"Q1_0","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","machine":"arm64","release":"25.6.0","run_id":"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","peak_run_id":"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","process_cpu_percent_avg":7.963578373856078,"process_rss_bytes_avg":6587746767.267236,"process_rss_bytes_peak":11685068800.0,"resource_cell_count":6,"sample_count":11302,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4949098496.0},"result_policy":"standard","runtime_key":"korgis-local","tags":["local","open-weight","bonsai","27b"]},"history":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-06T10:03:19.101787+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.5974025974025974,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-bonsai-27b-q10-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","execution_signature":"sha256:execution:80688f79e2c27746cabb9c8e268a98d63ee3d732cb5c73d70fc581a1958017fc","failure_count":0,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.475,"profile":"core","result_state":"CURRENT","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-06T09:40:29.476460+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.575,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-06T10:26:04.512560+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.948051948051948,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-06T09:38:49.212082+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-06T09:35:22.771531+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7833333333333333,"profile":"core","result_state":"CURRENT","run_id":"2026-10-core-v1-bonsai-27b-q10-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"model":{"artifact_format":"gguf","artifact_size_bytes":3803452480,"completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"bonsai-27b-q10","execution_signature":"sha256:execution:80688f79e2c27746cabb9c8e268a98d63ee3d732cb5c73d70fc581a1958017fc","family":"bonsai","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","parameters_b":27.0,"provider_key":"korgis","quantization":"Q1_0","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","run_status":"PARTIAL","runtime_key":"korgis-local"},"model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","schema_version":"2"},"sha256:model:fddf01c0845257f40bbb635a92054df8eb22dacd450b827a951962e906d6513b":{"current_cells":[{"artifact_format":"safetensors","artifact_size_bytes":416,"benchmark_signature":"sha256:benchmark:dca92e0e83499d6053d9585c0bdee4321b7ec9651563312492a659578feb94ee","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-20T20:00:36.317482+00:00","deployment":"local","effective_model_id":"limite-1b-violetto-mlx","execution_signature":"sha256:execution:f4c681453ad106c8c878d601ee4a0ee6335674e81abaf3f7ba5297ce35841db0","experiment_kind":"standard","failure_count":0,"family":"limite","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"gabe-rbo/limite-1b-violetto-mlx","model_key":"limite-1b-violetto-mlx","model_signature":"sha256:model:fddf01c0845257f40bbb635a92054df8eb22dacd450b827a951962e906d6513b","parameters_b":1.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.0,"profile":"smoke","provider_key":"korgis","quantization":null,"resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:f4c681453ad106c8c878d601ee4a0ee6335674e81abaf3f7ba5297ce35841db0","model_key":"limite-1b-violetto-mlx","process_cpu_percent_avg":54.47089950027608,"process_rss_bytes_avg":96917494.20591466,"process_rss_bytes_peak":102842368.0,"run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","sample_count":670,"sample_interval_ms":331.7098112721101,"sampling_error_count":0,"scope":"server_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":7568949248.0},"run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","runtime_key":"korgis-local","sample_count":10,"task_id":"mathematical-reasoning"}],"decision_summary":{"artifact_format":"safetensors","artifact_size_bytes":416,"current_run_ids":["d445eb9f-1c5a-4781-83be-43b6eeee3a44"],"deployment":"local","effective_model_id":"limite-1b-violetto-mlx","execution_count":1,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:f4c681453ad106c8c878d601ee4a0ee6335674e81abaf3f7ba5297ce35841db0","machine":"arm64","release":"25.6.0","run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"smoke","failure_count":10,"failure_rate":1.0,"family":"limite","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":21810.495604295284,"latency_p50_ms":21736.863624944817,"latency_p95_ms":22921.96444603032,"latest_completed_at_utc":"2026-10-20T20:00:36.317482+00:00","model_id":"gabe-rbo/limite-1b-violetto-mlx","model_key":"limite-1b-violetto-mlx","model_signature":"sha256:model:fddf01c0845257f40bbb635a92054df8eb22dacd450b827a951962e906d6513b","observed_case_count":10,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":null,"overall_quality_score":0.0,"parameters_b":1.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":10,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":10,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":1,"quality_coverage_complete":false,"quality_coverage_datasets":1,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":null,"resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:f4c681453ad106c8c878d601ee4a0ee6335674e81abaf3f7ba5297ce35841db0","machine":"arm64","release":"25.6.0","run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:f4c681453ad106c8c878d601ee4a0ee6335674e81abaf3f7ba5297ce35841db0","peak_run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","process_cpu_percent_avg":54.47089950027608,"process_rss_bytes_avg":96917494.20591466,"process_rss_bytes_peak":102842368.0,"resource_cell_count":1,"sample_count":670,"sampling_error_count":0,"scope":"server_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":7568949248.0},"result_policy":"standard","runtime_key":"korgis-local","tags":["local","open-weight","limite","violetto","mathematics","reasoning","small","1b","mlx","apple-silicon","korgis"]},"history":[{"benchmark_signature":"sha256:benchmark:dca92e0e83499d6053d9585c0bdee4321b7ec9651563312492a659578feb94ee","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-20T20:00:36.317482+00:00","execution_signature":"sha256:execution:f4c681453ad106c8c878d601ee4a0ee6335674e81abaf3f7ba5297ce35841db0","failure_count":0,"model_key":"limite-1b-violetto-mlx","model_signature":"sha256:model:fddf01c0845257f40bbb635a92054df8eb22dacd450b827a951962e906d6513b","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.0,"profile":"smoke","result_state":"CURRENT","run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","sample_count":10,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"model":{"artifact_format":"safetensors","artifact_size_bytes":416,"completed_at_utc":"2026-10-20T20:00:36.317482+00:00","deployment":"local","effective_model_id":"limite-1b-violetto-mlx","execution_signature":"sha256:execution:f4c681453ad106c8c878d601ee4a0ee6335674e81abaf3f7ba5297ce35841db0","family":"limite","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"gabe-rbo/limite-1b-violetto-mlx","model_key":"limite-1b-violetto-mlx","model_signature":"sha256:model:fddf01c0845257f40bbb635a92054df8eb22dacd450b827a951962e906d6513b","parameters_b":1.0,"provider_key":"korgis","quantization":null,"run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","run_status":"PARTIAL","runtime_key":"korgis-local"},"model_signature":"sha256:model:fddf01c0845257f40bbb635a92054df8eb22dacd450b827a951962e906d6513b","schema_version":"2"}},"overview":{"capabilities":["intent-classification","mathematical-reasoning","mathematical-reasoning-v2","oos-calibration","qa-abstention","structured-output"],"cells":[{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-06T10:03:19.101787+00:00","deployment":"local","effective_model_id":"bonsai-27b-q10","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.5974025974025974,"profile":"core","provider_key":"korgis","quantization":"Q1_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"bonsai-27b-q10","process_cpu_percent_avg":4.141997751956015,"process_rss_bytes_avg":5899303355.703257,"process_rss_bytes_peak":10491068416.0,"run_id":"2026-10-core-v1-bonsai-27b-q10-intent-classification-core","sample_count":4416,"sample_interval_ms":317.0853588508965,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4949098496.0},"run_id":"2026-10-core-v1-bonsai-27b-q10-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:718e761e4ba3410c289a5bba91a6474827caaa8da30e644b3df925ba4d16fbfc","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T15:04:59.981803+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"2e7b64ca38a2db66a9d8b36910d2b2ac5da0c7ce","model_id":"unsloth/qwen3.5-2b","model_key":"decisio-qwen3.5-2b-q4km","model_signature":"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.5974025974025974,"profile":"smoke","provider_key":"decisio","quantization":"Q4_K_M","resource_summary":null,"run_id":"16cd6c17-c9de-422d-8583-1a17959bdf6b","runtime_key":"decisio-local","sample_count":77,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T15:26:12.447493+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"2e7b64ca38a2db66a9d8b36910d2b2ac5da0c7ce","model_id":"unsloth/qwen3.5-2b","model_key":"decisio-qwen3.5-2b-q4km","model_signature":"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6038961038961039,"profile":"core","provider_key":"decisio","quantization":"Q4_K_M","resource_summary":null,"run_id":"1e48c427-4e75-4ace-a465-2aec7e2751cd","runtime_key":"decisio-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:718e761e4ba3410c289a5bba91a6474827caaa8da30e644b3df925ba4d16fbfc","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T15:11:29.046506+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"2e7b64ca38a2db66a9d8b36910d2b2ac5da0c7ce","model_id":"unsloth/qwen3.5-4b","model_key":"decisio-qwen3.5-4b-q4km","model_signature":"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6753246753246753,"profile":"smoke","provider_key":"decisio","quantization":"Q4_K_M","resource_summary":null,"run_id":"a97505cf-abc6-4782-bf99-38de2a4492b7","runtime_key":"decisio-local","sample_count":77,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T15:33:33.988827+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"2e7b64ca38a2db66a9d8b36910d2b2ac5da0c7ce","model_id":"unsloth/qwen3.5-4b","model_key":"decisio-qwen3.5-4b-q4km","model_signature":"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.7012987012987013,"profile":"core","provider_key":"decisio","quantization":"Q4_K_M","resource_summary":null,"run_id":"b906e312-aa49-4397-b60e-a1c3ad6a8f87","runtime_key":"decisio-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:718e761e4ba3410c289a5bba91a6474827caaa8da30e644b3df925ba4d16fbfc","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T16:17:35.400024+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"0065acc5513f65c3e8bbdf6899cf55e286dfe938","model_id":"qwen/qwen3.5-9b","model_key":"decisio-qwen3.5-9b-q4km","model_signature":"sha256:model:ae0099eeb8925bc828c92072ba2f486986899854d2d3de76e5689cd383c783e3","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6623376623376623,"profile":"smoke","provider_key":"decisio","quantization":"Q4_K_M","resource_summary":null,"run_id":"e9c4ec37-433f-4b06-b397-9d9b3a4f7480","runtime_key":"decisio-local","sample_count":77,"task_id":"intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T13:59:58.574010+00:00","deployment":"api","effective_model_id":"gemini-3.5-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.5","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.0,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core","runtime_key":"gemini-api","sample_count":154,"task_id":"intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T13:39:20.878424+00:00","deployment":"api","effective_model_id":"gemini-3.8-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.8","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.8-flash","model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.0,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core","runtime_key":"gemini-api","sample_count":154,"task_id":"intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-06T12:22:05.300981+00:00","deployment":"api","effective_model_id":"gpt-5.6-luna","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"11b00f55c776ec89ddee7cd2dd1898797dde4055","model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.8246753246753247,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-intent-classification-core","runtime_key":"openai-api","sample_count":154,"task_id":"intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T13:18:34.815598+00:00","deployment":"api","effective_model_id":"gpt-6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","experiment_kind":"standard","failure_count":0,"family":"gpt-6","generation_policy":{},"git_commit":"7f796c10d99e934da6417ea54405fbf9352ad399","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.7987012987012987,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"gpt-6-luna-core-gpt-6-luna-intent-classification-core","runtime_key":"openai-api","sample_count":154,"task_id":"intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-06T12:50:59.652242+00:00","deployment":"api","effective_model_id":"jev-1.13.0","execution_signature":"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3","experiment_kind":"standard","failure_count":0,"family":"jev","generation_policy":{},"git_commit":"72697127df45f8a3abc98b0b7bdf6fd53a0b5c98","model_id":"jev-1.13.0","model_key":"jev-1.13.0","model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.7467532467532467,"profile":"core","provider_key":"typesafe","quantization":null,"resource_summary":null,"run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","runtime_key":"typesafe-api","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":10947100928,"benchmark_signature":"sha256:consolidated:42ab94943101512840504e431240a2dc1e9c2c51d6e055498c7b6b44c803f515","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-22T13:12:52.242438+00:00","deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","experiment_kind":"consolidated","failure_count":0,"family":"lfm2.5","generation_policy":{"generation_configs":[{"extra":{},"max_output_tokens":2048,"seed":42,"stop":[],"temperature":0.0},{"extra":{},"max_output_tokens":4096,"seed":42,"stop":[],"temperature":0.0}],"mixed_generation":true},"git_commit":null,"model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.7012987012987013,"profile":"core","provider_key":"korgis","quantization":"UD-Q3_K_XL","resource_summary":null,"run_id":"lfm25-consolidated-latest","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T12:03:20.624151+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.5584415584415584,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b","process_cpu_percent_avg":5.91751681481258,"process_rss_bytes_avg":5777497087.999999,"process_rss_bytes_peak":10717691904.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-intent-classification-core","sample_count":1252,"sample_interval_ms":293.3800480326384,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6120767488.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T13:24:18.004156+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b-q8","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.551948051948052,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b-q8","process_cpu_percent_avg":4.532544993975123,"process_rss_bytes_avg":5542900553.420978,"process_rss_bytes_peak":9062940672.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core","sample_count":1798,"sample_interval_ms":315.6007340511925,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5441241088.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T13:37:29.943647+00:00","deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.2792207792207792,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-0.8b-q4km","process_cpu_percent_avg":12.95914646037019,"process_rss_bytes_avg":4417102107.705627,"process_rss_bytes_peak":7946436608.0,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core","sample_count":462,"sample_interval_ms":199.48604018792173,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":10223632384.0},"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":3775709216,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T11:24:59.796058+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-bf16","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","parameters_b":2.0,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.577922077922078,"profile":"core","provider_key":"korgis","quantization":"BF16","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","model_key":"qwen3.5-2b-bf16","process_cpu_percent_avg":5.488711560059013,"process_rss_bytes_avg":5427256115.200001,"process_rss_bytes_peak":7206486016.0,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core","sample_count":618,"sample_interval_ms":285.9658953851292,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5988106240.0},"run_id":"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T11:25:37.855179+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.577922077922078,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-2b-q4km","process_cpu_percent_avg":9.421328984702535,"process_rss_bytes_avg":2155211762.7012987,"process_rss_bytes_peak":3131637760.0,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core","sample_count":616,"sample_interval_ms":260.0283379352414,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6455427072.0},"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T16:36:39.120802+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"cb061291446b8dbd79f524a4fefcd4a4406e2a0d","model_id":"unsloth/qwen3.5-4b","model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6688311688311688,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-4b-q4km","process_cpu_percent_avg":6.008969494809468,"process_rss_bytes_avg":5011243553.1201,"process_rss_bytes_peak":7911129088.0,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":1067,"sample_interval_ms":281.19401557671836,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5343313920.0},"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":4482403488,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T11:37:13.963536+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q8","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","parameters_b":4.0,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6493506493506493,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","model_key":"qwen3.5-4b-q8","process_cpu_percent_avg":5.113588871923991,"process_rss_bytes_avg":7039335235.285094,"process_rss_bytes_peak":10390700032.0,"run_id":"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core","sample_count":1041,"sample_interval_ms":311.43875954839615,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5581979648.0},"run_id":"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T12:46:44.225239+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6948051948051948,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-9b-q4km","process_cpu_percent_avg":3.9138417761512048,"process_rss_bytes_avg":6965862817.821014,"process_rss_bytes_peak":10559422464.0,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core","sample_count":1620,"sample_interval_ms":303.4210088575144,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5489934336.0},"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":4121781472,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T11:55:33.171703+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-ud-q2kxl","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"qwen/qwen3.5-9b-ud-q2kxl","model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","parameters_b":9.0,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6883116883116883,"profile":"core","provider_key":"korgis","quantization":"UD-Q2_K_XL","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","model_key":"qwen3.5-9b-ud-q2kxl","process_cpu_percent_avg":3.288803542984178,"process_rss_bytes_avg":6954068738.250033,"process_rss_bytes_peak":11224416256.0,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","sample_count":1539,"sample_interval_ms":314.5480896344338,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5547704320.0},"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":6192222208,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-06T14:37:14.912068+00:00","deployment":"local","effective_model_id":"qwen3.8-27b-iq1s","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","experiment_kind":"standard","failure_count":0,"family":"qwen3.8","generation_policy":{},"git_commit":"640138c3429882a42797d28074f98d3fea98a63b","model_id":"Qwen/Qwen3.8-27B","model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","parameters_b":27.0,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.5844155844155844,"profile":"core","provider_key":"korgis","quantization":"UD-IQ1_S","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","model_key":"qwen3.8-27b-iq1s","process_cpu_percent_avg":3.5068517339810183,"process_rss_bytes_avg":8038410776.168479,"process_rss_bytes_peak":12158550016.0,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","sample_count":4560,"sample_interval_ms":309.607195745226,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5166546944.0},"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-06T09:08:01.471590+00:00","deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"xhtoken/spark-x2.5-4b","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","parameters_b":null,"practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6493506493506493,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"spark-x2.5-4b-q4km","process_cpu_percent_avg":5.594450252441077,"process_rss_bytes_avg":3797578739.069748,"process_rss_bytes_peak":6051266560.0,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core","sample_count":1308,"sample_interval_ms":310.7691591083291,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5891014656.0},"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core","runtime_key":"korgis-local","sample_count":154,"task_id":"intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":3803452480,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"bonsai-27b-q10","execution_signature":"sha256:execution:80688f79e2c27746cabb9c8e268a98d63ee3d732cb5c73d70fc581a1958017fc","experiment_kind":"standard","failure_count":0,"family":"bonsai","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","parameters_b":27.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.475,"profile":"core","provider_key":"korgis","quantization":"Q1_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:80688f79e2c27746cabb9c8e268a98d63ee3d732cb5c73d70fc581a1958017fc","model_key":"bonsai-27b-q10","process_cpu_percent_avg":10.004219049002119,"process_rss_bytes_avg":7178716753.609695,"process_rss_bytes_peak":9109520384.0,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":342,"sample_interval_ms":299.5298028563516,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5797576704.0},"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"api","effective_model_id":"gemini-3.5-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.5","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.125,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"gemini-api","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T15:34:31.093143+00:00","deployment":"api","effective_model_id":"gpt-5.6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","experiment_kind":"standard","failure_count":0,"family":"gpt-5.6","generation_policy":{},"git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.675,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","runtime_key":"openai-api","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T15:34:31.093143+00:00","deployment":"api","effective_model_id":"gpt-6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","experiment_kind":"standard","failure_count":0,"family":"gpt-6","generation_policy":{},"git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.7,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","runtime_key":"openai-api","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":10947100928,"benchmark_signature":"sha256:consolidated:20f65d6394555078bc9dce0c1393e10d963aecb84ce9344ece12046a78d42de3","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-22T13:12:52.242438+00:00","deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","experiment_kind":"consolidated","failure_count":0,"family":"lfm2.5","generation_policy":{"generation_configs":[{"extra":{},"max_output_tokens":2048,"seed":42,"stop":[],"temperature":0.0},{"extra":{},"max_output_tokens":4096,"seed":42,"stop":[],"temperature":0.0}],"mixed_generation":true},"git_commit":null,"model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.95,"profile":"core","provider_key":"korgis","quantization":"UD-Q3_K_XL","resource_summary":null,"run_id":"lfm25-consolidated-latest","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:27:15.751735+00:00","deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","experiment_kind":"standard","failure_count":0,"family":"limite","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","parameters_b":1.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.9,"profile":"core","provider_key":"transformers","quantization":null,"resource_summary":null,"run_id":"16a842b0-5dcc-47b5-8808-850b2ad1ac10","runtime_key":"transformers-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:382be54f0e8c250212fa1d53656a34eb22335c491d7aa5b1ba59e927a0c58dea","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T14:59:31.170321+00:00","deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","experiment_kind":"standard","failure_count":0,"family":"limite","generation_policy":{},"git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65","parameters_b":1.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.7222222222222222,"profile":"extended","provider_key":"transformers","quantization":null,"resource_summary":null,"run_id":"2533bf73-b012-4d12-a6b6-9b6e4334dae2","runtime_key":"transformers-local","sample_count":72,"task_id":"mathematical-reasoning"},{"artifact_format":"safetensors","artifact_size_bytes":416,"benchmark_signature":"sha256:benchmark:dca92e0e83499d6053d9585c0bdee4321b7ec9651563312492a659578feb94ee","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-20T20:00:36.317482+00:00","deployment":"local","effective_model_id":"limite-1b-violetto-mlx","execution_signature":"sha256:execution:f4c681453ad106c8c878d601ee4a0ee6335674e81abaf3f7ba5297ce35841db0","experiment_kind":"standard","failure_count":0,"family":"limite","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"gabe-rbo/limite-1b-violetto-mlx","model_key":"limite-1b-violetto-mlx","model_signature":"sha256:model:fddf01c0845257f40bbb635a92054df8eb22dacd450b827a951962e906d6513b","parameters_b":1.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.0,"profile":"smoke","provider_key":"korgis","quantization":null,"resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:f4c681453ad106c8c878d601ee4a0ee6335674e81abaf3f7ba5297ce35841db0","model_key":"limite-1b-violetto-mlx","process_cpu_percent_avg":54.47089950027608,"process_rss_bytes_avg":96917494.20591466,"process_rss_bytes_peak":102842368.0,"run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","sample_count":670,"sample_interval_ms":331.7098112721101,"sampling_error_count":0,"scope":"server_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":7568949248.0},"run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","runtime_key":"korgis-local","sample_count":10,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":2837072896,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T15:45:29.838316+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b","execution_signature":"sha256:execution:0caf2fff221781464037ba78ff0ed5a8158d632ab0e66481fc8df733dd8785c6","experiment_kind":"standard","failure_count":0,"family":"nemotron-nano","generation_policy":{},"git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","parameters_b":4.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.5,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:0caf2fff221781464037ba78ff0ed5a8158d632ab0e66481fc8df733dd8785c6","model_key":"nemotron-nano-4b","process_cpu_percent_avg":10.014690415304136,"process_rss_bytes_avg":4240616618.6666665,"process_rss_bytes_peak":4889542656.0,"run_id":"3b0292d4-8f8f-4843-900c-946c32476964","sample_count":153,"sample_interval_ms":267.2538513759112,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5000953856.0},"run_id":"3b0292d4-8f8f-4843-900c-946c32476964","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":4233677952,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b-q8","execution_signature":"sha256:execution:89a74e35df26090fe2c6adb16b01e32373fbcdad6966a1b23cc9b88ba35a5170","experiment_kind":"standard","failure_count":0,"family":"nemotron-nano","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","parameters_b":4.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.475,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:89a74e35df26090fe2c6adb16b01e32373fbcdad6966a1b23cc9b88ba35a5170","model_key":"nemotron-nano-4b-q8","process_cpu_percent_avg":8.818030339465103,"process_rss_bytes_avg":7560277067.093333,"process_rss_bytes_peak":10620682240.0,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":152,"sample_interval_ms":260.26064477991895,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":10035691520.0},"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":532517120,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","execution_signature":"sha256:execution:c58f103ecf1090aab4510c6708cd9446fefc4137c61ae9ca33da8cc1c5d3faf8","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","parameters_b":0.8,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.1,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:c58f103ecf1090aab4510c6708cd9446fefc4137c61ae9ca33da8cc1c5d3faf8","model_key":"qwen3.5-0.8b-q4km","process_cpu_percent_avg":15.698688370479838,"process_rss_bytes_avg":1544970035.2,"process_rss_bytes_peak":2028224512.0,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":80,"sample_interval_ms":263.3856863947585,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5715886080.0},"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":3775709216,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-bf16","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","parameters_b":2.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.25,"profile":"core","provider_key":"korgis","quantization":"BF16","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","model_key":"qwen3.5-2b-bf16","process_cpu_percent_avg":8.718440372188164,"process_rss_bytes_avg":4446060373.333334,"process_rss_bytes_peak":4834885632.0,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":121,"sample_interval_ms":268.43950018325506,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5017370624.0},"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":1280835840,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T15:44:53.050810+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:8190c8f3bb1f7b51ef597577c913ceae05f8cb7998a7d5221e848f67071166c9","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","parameters_b":2.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.25,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:8190c8f3bb1f7b51ef597577c913ceae05f8cb7998a7d5221e848f67071166c9","model_key":"qwen3.5-2b-q4km","process_cpu_percent_avg":8.953256029208966,"process_rss_bytes_avg":2160430830.9333334,"process_rss_bytes_peak":2483290112.0,"run_id":"9fb9f932-1f29-4646-b337-f33476d841ba","sample_count":94,"sample_interval_ms":267.4467081844341,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5225758720.0},"run_id":"9fb9f932-1f29-4646-b337-f33476d841ba","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":2740937888,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:d6ef7d6b6d33f70d69c5d1df4568941686a610cda0bde994801472a902500dd4","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"unsloth/qwen3.5-4b","model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","parameters_b":4.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.4,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:d6ef7d6b6d33f70d69c5d1df4568941686a610cda0bde994801472a902500dd4","model_key":"qwen3.5-4b-q4km","process_cpu_percent_avg":10.67374362291546,"process_rss_bytes_avg":4568710143.999999,"process_rss_bytes_peak":5363400704.0,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":129,"sample_interval_ms":303.8685138652605,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5199380480.0},"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":4482403488,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q8","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","parameters_b":4.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.475,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","model_key":"qwen3.5-4b-q8","process_cpu_percent_avg":10.86561479591514,"process_rss_bytes_avg":5559348804.266666,"process_rss_bytes_peak":6319587328.0,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":146,"sample_interval_ms":295.3429861280407,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4841750528.0},"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":5627044256,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:9d728f687c9bb0f47a2c5a8c0e61b6da5a7e2833ad1afcffb18c7b4eb2767886","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","parameters_b":9.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.5,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:9d728f687c9bb0f47a2c5a8c0e61b6da5a7e2833ad1afcffb18c7b4eb2767886","model_key":"qwen3.5-9b-q4km","process_cpu_percent_avg":7.04412305796482,"process_rss_bytes_avg":8143566028.8,"process_rss_bytes_peak":9486303232.0,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":176,"sample_interval_ms":291.7408815034529,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":9481781248.0},"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":2600224352,"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T15:45:57.052259+00:00","deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_signature":"sha256:execution:d26122294632fb1ef7e62445eece13d61030562e5a3d7a64b60606c3c12ff3be","experiment_kind":"standard","failure_count":0,"family":"spark-x2.5","generation_policy":{},"git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","model_id":"XHToken/Spark-X2.5-4B","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:8dcf43803761fd5570f637e166b655565b6c638e51a6a96848f58f6e6699a734","parameters_b":4.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.3,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:d26122294632fb1ef7e62445eece13d61030562e5a3d7a64b60606c3c12ff3be","model_key":"spark-x2.5-4b-q4km","process_cpu_percent_avg":6.338877896022156,"process_rss_bytes_avg":3725684701.8666673,"process_rss_bytes_peak":4221812736.0,"run_id":"e68754ab-25e6-4449-b06a-66dbc311d9e5","sample_count":122,"sample_interval_ms":266.60380743026815,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":8769241088.0},"run_id":"e68754ab-25e6-4449-b06a-66dbc311d9e5","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-06T09:40:29.476460+00:00","deployment":"local","effective_model_id":"bonsai-27b-q10","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.575,"profile":"core","provider_key":"korgis","quantization":"Q1_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"bonsai-27b-q10","process_cpu_percent_avg":10.398893667073832,"process_rss_bytes_avg":7636308908.292063,"process_rss_bytes_peak":11685068800.0,"run_id":"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","sample_count":340,"sample_interval_ms":307.1656581635226,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5910167552.0},"run_id":"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T13:59:36.375684+00:00","deployment":"api","effective_model_id":"gemini-3.5-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.5","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.0,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core","runtime_key":"gemini-api","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T13:38:58.640898+00:00","deployment":"api","effective_model_id":"gemini-3.8-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.8","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.8-flash","model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.0,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core","runtime_key":"gemini-api","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-06T12:17:36.689339+00:00","deployment":"api","effective_model_id":"gpt-5.6-luna","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.45,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core","runtime_key":"openai-api","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T13:13:20.819089+00:00","deployment":"api","effective_model_id":"gpt-6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","experiment_kind":"standard","failure_count":0,"family":"gpt-6","generation_policy":{},"git_commit":"7f796c10d99e934da6417ea54405fbf9352ad399","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.425,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core","runtime_key":"openai-api","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":10947100928,"benchmark_signature":"sha256:benchmark:600e8b1d94c1e2bc55bd8802918ce0c6f18dece6d8a9c02247a530fc7b015409","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-22T14:07:00.510877+00:00","deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_signature":"sha256:execution:503d302009e21bb18588b87ac3897f9112994a479905bbca872384efa4360298","experiment_kind":"standard","failure_count":0,"family":"lfm2.5","generation_policy":{"max_output_tokens":4096},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.775,"profile":"core","provider_key":"korgis","quantization":"UD-Q3_K_XL","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:503d302009e21bb18588b87ac3897f9112994a479905bbca872384efa4360298","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","process_cpu_percent_avg":10.29314878109045,"process_rss_bytes_avg":1417688467.7424607,"process_rss_bytes_peak":1463648256.0,"run_id":"lfm25-math-v2-core","sample_count":1316,"sample_interval_ms":313.23082923487743,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4090920960.0},"run_id":"lfm25-math-v2-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-17T21:24:56.752644+00:00","deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","experiment_kind":"standard","failure_count":0,"family":"limite","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","parameters_b":1.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.8,"profile":"core","provider_key":"transformers","quantization":null,"resource_summary":null,"run_id":"8eef4bab-617a-4638-9e51-53e72034e821","runtime_key":"transformers-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:bff2038c9a70a239e320793c0c579d02b35645dd0f722e6fd8e0c6a9b30b4b07","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T13:35:21.786434+00:00","deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","experiment_kind":"standard","failure_count":0,"family":"limite","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65","parameters_b":1.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.8333333333333334,"profile":"smoke","provider_key":"transformers","quantization":null,"resource_summary":null,"run_id":"cb837fcf-570f-4a89-bbbd-cded7a3c335b","runtime_key":"transformers-local","sample_count":12,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T12:48:39.598492+00:00","deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:ff14676ac39e23af360cd1209491ef1c14e3dab9a3198b48054de5c6de699fdb","experiment_kind":"standard","failure_count":0,"family":"limite","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:3fe1bfff5bb2c952f96bb799970eb76d9db17f9f880bc3326f0bdfbdd51577cc","parameters_b":1.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.0,"profile":"core","provider_key":"vllm","quantization":null,"resource_summary":null,"run_id":"5623e5c1-213d-447a-88e9-ac51a69a92fa","runtime_key":"vllm-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T13:46:02.371339+00:00","deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","experiment_kind":"standard","failure_count":0,"family":"limite","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65","parameters_b":1.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.725,"profile":"core","provider_key":"transformers","quantization":null,"resource_summary":null,"run_id":"954053b1-6f87-4196-bb2a-028404b6fef1","runtime_key":"transformers-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T11:57:43.663487+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.575,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b","process_cpu_percent_avg":11.038100201393934,"process_rss_bytes_avg":7133634082.133334,"process_rss_bytes_peak":11070996480.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core","sample_count":143,"sample_interval_ms":255.84691734608595,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":11042144256.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T13:15:22.380707+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b-q8","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.6,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b-q8","process_cpu_percent_avg":8.601215146103618,"process_rss_bytes_avg":7892413781.333334,"process_rss_bytes_peak":11438456832.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","sample_count":158,"sample_interval_ms":252.1124605016667,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":7386234880.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T13:36:13.768283+00:00","deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.2,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-0.8b-q4km","process_cpu_percent_avg":18.636242311523596,"process_rss_bytes_avg":1745210777.6,"process_rss_bytes_peak":2670067712.0,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core","sample_count":81,"sample_interval_ms":213.9264905885284,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":15259779072.0},"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":3775709216,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T11:22:33.733080+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-bf16","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","parameters_b":2.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.375,"profile":"core","provider_key":"korgis","quantization":"BF16","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","model_key":"qwen3.5-2b-bf16","process_cpu_percent_avg":6.761276471009742,"process_rss_bytes_avg":5075235362.133334,"process_rss_bytes_peak":5976489984.0,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core","sample_count":120,"sample_interval_ms":240.70379737386247,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":10141270016.0},"run_id":"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T11:23:17.557837+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.325,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-2b-q4km","process_cpu_percent_avg":12.866597281256125,"process_rss_bytes_avg":2496888695.466666,"process_rss_bytes_peak":3418226688.0,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core","sample_count":104,"sample_interval_ms":243.8169525759804,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":8507686912.0},"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T16:36:39.120802+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"cb061291446b8dbd79f524a4fefcd4a4406e2a0d","model_id":"unsloth/qwen3.5-4b","model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.475,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-4b-q4km","process_cpu_percent_avg":12.689180073192913,"process_rss_bytes_avg":5970077900.800001,"process_rss_bytes_peak":7198195712.0,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":134,"sample_interval_ms":272.2716644006747,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":8184053760.0},"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":4482403488,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T11:32:22.918386+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q8","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","parameters_b":4.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.5,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","model_key":"qwen3.5-4b-q8","process_cpu_percent_avg":6.982037738154942,"process_rss_bytes_avg":7516327697.066668,"process_rss_bytes_peak":9962160128.0,"run_id":"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core","sample_count":132,"sample_interval_ms":291.2576532937237,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":11326062592.0},"run_id":"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T12:39:03.216607+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.675,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-9b-q4km","process_cpu_percent_avg":6.526783901766963,"process_rss_bytes_avg":8611465482.24,"process_rss_bytes_peak":11088855040.0,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core","sample_count":175,"sample_interval_ms":275.92921112542774,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":11636277248.0},"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":4121781472,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T11:48:03.359698+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-ud-q2kxl","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"qwen/qwen3.5-9b-ud-q2kxl","model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","parameters_b":9.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.55,"profile":"core","provider_key":"korgis","quantization":"UD-Q2_K_XL","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","model_key":"qwen3.5-9b-ud-q2kxl","process_cpu_percent_avg":4.913551711729864,"process_rss_bytes_avg":7164752896.0,"process_rss_bytes_peak":9602973696.0,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core","sample_count":164,"sample_interval_ms":288.91353003637056,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":12627738624.0},"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":6192222208,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-06T14:37:12.067024+00:00","deployment":"local","effective_model_id":"qwen3.8-27b-iq1s","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","experiment_kind":"standard","failure_count":0,"family":"qwen3.8","generation_policy":{},"git_commit":"640138c3429882a42797d28074f98d3fea98a63b","model_id":"Qwen/Qwen3.8-27B","model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","parameters_b":27.0,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.25,"profile":"core","provider_key":"korgis","quantization":"UD-IQ1_S","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","model_key":"qwen3.8-27b-iq1s","process_cpu_percent_avg":7.113533739353301,"process_rss_bytes_avg":9138898947.755379,"process_rss_bytes_peak":12174065664.0,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","sample_count":717,"sample_interval_ms":297.72640370809415,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5485953024.0},"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-06T09:01:45.488881+00:00","deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"xhtoken/spark-x2.5-4b","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","parameters_b":null,"practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.45,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"spark-x2.5-4b-q4km","process_cpu_percent_avg":8.149214468504207,"process_rss_bytes_avg":3498299289.6,"process_rss_bytes_peak":3887923200.0,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core","sample_count":147,"sample_interval_ms":267.7047739123735,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":7349649408.0},"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core","runtime_key":"korgis-local","sample_count":40,"task_id":"mathematical-reasoning"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-06T10:26:04.512560+00:00","deployment":"local","effective_model_id":"bonsai-27b-q10","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.948051948051948,"profile":"core","provider_key":"korgis","quantization":"Q1_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"bonsai-27b-q10","process_cpu_percent_avg":3.871693631322875,"process_rss_bytes_avg":5867857169.246841,"process_rss_bytes_peak":9145516032.0,"run_id":"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core","sample_count":4524,"sample_interval_ms":309.6376288065501,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5604933632.0},"run_id":"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T14:00:21.676846+00:00","deployment":"api","effective_model_id":"gemini-3.5-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.5","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.5,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core","runtime_key":"gemini-api","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T13:39:43.452981+00:00","deployment":"api","effective_model_id":"gemini-3.8-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.8","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.8-flash","model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.5,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core","runtime_key":"gemini-api","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-06T12:26:04.935243+00:00","deployment":"api","effective_model_id":"gpt-5.6-luna","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"11b00f55c776ec89ddee7cd2dd1898797dde4055","model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":1.0,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core","runtime_key":"openai-api","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T13:25:20.495814+00:00","deployment":"api","effective_model_id":"gpt-6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","experiment_kind":"standard","failure_count":0,"family":"gpt-6","generation_policy":{},"git_commit":"7f796c10d99e934da6417ea54405fbf9352ad399","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.9935064935064936,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"gpt-6-luna-core-gpt-6-luna-oos-calibration-core","runtime_key":"openai-api","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-06T12:51:50.754874+00:00","deployment":"api","effective_model_id":"jev-1.13.0","execution_signature":"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3","experiment_kind":"standard","failure_count":0,"family":"jev","generation_policy":{},"git_commit":"72697127df45f8a3abc98b0b7bdf6fd53a0b5c98","model_id":"jev-1.13.0","model_key":"jev-1.13.0","model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.974025974025974,"profile":"core","provider_key":"typesafe","quantization":null,"resource_summary":null,"run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","runtime_key":"typesafe-api","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":10947100928,"benchmark_signature":"sha256:consolidated:9468d75aebe9e805cefc3643f61a9e5fe6127c7a4e3a987608b30841b9bc9fa0","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-22T13:12:52.242438+00:00","deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","experiment_kind":"consolidated","failure_count":0,"family":"lfm2.5","generation_policy":{"generation_configs":[{"extra":{},"max_output_tokens":2048,"seed":42,"stop":[],"temperature":0.0},{"extra":{},"max_output_tokens":4096,"seed":42,"stop":[],"temperature":0.0}],"mixed_generation":true},"git_commit":null,"model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.9545454545454546,"profile":"core","provider_key":"korgis","quantization":"UD-Q3_K_XL","resource_summary":null,"run_id":"lfm25-consolidated-latest","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T12:10:00.371141+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b","process_cpu_percent_avg":5.490567532375733,"process_rss_bytes_avg":5130473995.082253,"process_rss_bytes_peak":10141679616.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core","sample_count":1444,"sample_interval_ms":296.66646442619066,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6067978240.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T13:32:50.472664+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b-q8","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b-q8","process_cpu_percent_avg":5.151403915249351,"process_rss_bytes_avg":5654897500.709869,"process_rss_bytes_peak":8370143232.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core","sample_count":1747,"sample_interval_ms":306.91440564565875,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5491818496.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T13:39:04.217888+00:00","deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.5,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-0.8b-q4km","process_cpu_percent_avg":16.11579774911694,"process_rss_bytes_avg":5182799526.2337675,"process_rss_bytes_peak":9496838144.0,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","sample_count":463,"sample_interval_ms":247.8098196266941,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":9132261376.0},"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":3775709216,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T11:28:16.241563+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-bf16","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","parameters_b":2.0,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.5,"profile":"core","provider_key":"korgis","quantization":"BF16","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","model_key":"qwen3.5-2b-bf16","process_cpu_percent_avg":5.117048169336818,"process_rss_bytes_avg":5994336968.810392,"process_rss_bytes_peak":9291366400.0,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","sample_count":770,"sample_interval_ms":293.5350881957497,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5950177280.0},"run_id":"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T11:28:16.098461+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.5,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-2b-q4km","process_cpu_percent_avg":8.346543248013683,"process_rss_bytes_avg":3164761335.3558445,"process_rss_bytes_peak":6107152384.0,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","sample_count":621,"sample_interval_ms":291.97748509630554,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6536364032.0},"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T16:36:39.120802+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"cb061291446b8dbd79f524a4fefcd4a4406e2a0d","model_id":"unsloth/qwen3.5-4b","model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.974025974025974,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-4b-q4km","process_cpu_percent_avg":6.406836413976952,"process_rss_bytes_avg":6021487127.958771,"process_rss_bytes_peak":10610851840.0,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":1193,"sample_interval_ms":286.7970046134617,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5619220480.0},"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":4482403488,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T11:43:07.482305+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q8","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","parameters_b":4.0,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.9805194805194806,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","model_key":"qwen3.5-4b-q8","process_cpu_percent_avg":5.610485336009209,"process_rss_bytes_avg":7096272805.442177,"process_rss_bytes_peak":11637391360.0,"run_id":"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","sample_count":1211,"sample_interval_ms":319.76690310093494,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5589811200.0},"run_id":"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T13:01:17.635848+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.974025974025974,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-9b-q4km","process_cpu_percent_avg":2.683081412939199,"process_rss_bytes_avg":6844874999.319381,"process_rss_bytes_peak":10287939584.0,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core","sample_count":2848,"sample_interval_ms":314.6654996565356,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5142609920.0},"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":4121781472,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T12:04:10.873473+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-ud-q2kxl","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"qwen/qwen3.5-9b-ud-q2kxl","model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","parameters_b":9.0,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"profile":"core","provider_key":"korgis","quantization":"UD-Q2_K_XL","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","model_key":"qwen3.5-9b-ud-q2kxl","process_cpu_percent_avg":3.489486960210178,"process_rss_bytes_avg":6127369651.484418,"process_rss_bytes_peak":9636659200.0,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core","sample_count":1713,"sample_interval_ms":322.34991046526653,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5265096704.0},"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":6192222208,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-06T14:53:01.287222+00:00","deployment":"local","effective_model_id":"qwen3.8-27b-iq1s","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","experiment_kind":"standard","failure_count":0,"family":"qwen3.8","generation_policy":{},"git_commit":"640138c3429882a42797d28074f98d3fea98a63b","model_id":"Qwen/Qwen3.8-27B","model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","parameters_b":27.0,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.8181818181818182,"profile":"core","provider_key":"korgis","quantization":"UD-IQ1_S","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","model_key":"qwen3.8-27b-iq1s","process_cpu_percent_avg":3.3561674118938543,"process_rss_bytes_avg":8080732806.054676,"process_rss_bytes_peak":11618025472.0,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","sample_count":6492,"sample_interval_ms":307.9710546548535,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5336121344.0},"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-06T09:14:39.475593+00:00","deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"xhtoken/spark-x2.5-4b","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","parameters_b":null,"practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"spark-x2.5-4b-q4km","process_cpu_percent_avg":5.353823389553569,"process_rss_bytes_avg":4696997069.724959,"process_rss_bytes_peak":7862026240.0,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","sample_count":1416,"sample_interval_ms":302.2578173696654,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6040551424.0},"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","runtime_key":"korgis-local","sample_count":154,"task_id":"calibrated-intent-classification"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-06T09:38:49.212082+00:00","deployment":"local","effective_model_id":"bonsai-27b-q10","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"profile":"core","provider_key":"korgis","quantization":"Q1_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"bonsai-27b-q10","process_cpu_percent_avg":9.839476147142088,"process_rss_bytes_avg":6890131131.833357,"process_rss_bytes_peak":10222600192.0,"run_id":"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core","sample_count":690,"sample_interval_ms":311.07264410993747,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5791531008.0},"run_id":"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T13:59:29.849450+00:00","deployment":"api","effective_model_id":"gemini-3.5-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.5","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.016666666666666666,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core","runtime_key":"gemini-api","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T13:38:52.091017+00:00","deployment":"api","effective_model_id":"gemini-3.8-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.8","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.8-flash","model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.0,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core","runtime_key":"gemini-api","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-06T12:16:26.396465+00:00","deployment":"api","effective_model_id":"gpt-5.6-luna","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core","runtime_key":"openai-api","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T13:12:07.696210+00:00","deployment":"api","effective_model_id":"gpt-6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","experiment_kind":"standard","failure_count":0,"family":"gpt-6","generation_policy":{},"git_commit":"7f796c10d99e934da6417ea54405fbf9352ad399","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.95,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"gpt-6-luna-core-gpt-6-luna-qa-abstention-core","runtime_key":"openai-api","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":10947100928,"benchmark_signature":"sha256:consolidated:b832fe612fcfe138845fe193a557c686f68ecb2f59d1b3f4784e653e9ec82e5c","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-22T13:12:52.242438+00:00","deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","experiment_kind":"consolidated","failure_count":0,"family":"lfm2.5","generation_policy":{"generation_configs":[{"extra":{},"max_output_tokens":2048,"seed":42,"stop":[],"temperature":0.0},{"extra":{},"max_output_tokens":4096,"seed":42,"stop":[],"temperature":0.0}],"mixed_generation":true},"git_commit":null,"model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"profile":"core","provider_key":"korgis","quantization":"UD-Q3_K_XL","resource_summary":null,"run_id":"lfm25-consolidated-latest","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:1fa41ef05579230896c951f4e01da57ccf719647ad6f8235008243556409a7cb","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-20T10:03:44.604674+00:00","deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","experiment_kind":"standard","failure_count":0,"family":"limite","generation_policy":{},"git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","parameters_b":1.0,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.0,"profile":"smoke","provider_key":"transformers","quantization":null,"resource_summary":null,"run_id":"b6b4123a-1d20-4c39-8644-ea00c1e5893b","runtime_key":"transformers-local","sample_count":12,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T11:57:09.469183+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b","process_cpu_percent_avg":9.533096212147692,"process_rss_bytes_avg":7103316869.12,"process_rss_bytes_peak":11089903616.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","sample_count":264,"sample_interval_ms":273.58964924757555,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6888439808.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T13:14:43.428857+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b-q8","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b-q8","process_cpu_percent_avg":7.166483605707706,"process_rss_bytes_avg":6463349882.22984,"process_rss_bytes_peak":10168467456.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core","sample_count":359,"sample_interval_ms":297.72169937988014,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5578047488.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T13:35:56.623493+00:00","deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.48333333333333334,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-0.8b-q4km","process_cpu_percent_avg":16.023245275126758,"process_rss_bytes_avg":2421507094.7555556,"process_rss_bytes_peak":3926736896.0,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core","sample_count":126,"sample_interval_ms":259.9318482166079,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":14612348928.0},"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":3775709216,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T11:22:09.705119+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-bf16","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","parameters_b":2.0,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.6333333333333333,"profile":"core","provider_key":"korgis","quantization":"BF16","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","model_key":"qwen3.5-2b-bf16","process_cpu_percent_avg":5.5345916105632815,"process_rss_bytes_avg":5136314017.564444,"process_rss_bytes_peak":5894602752.0,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core","sample_count":226,"sample_interval_ms":279.53779863981055,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":7390167040.0},"run_id":"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T11:22:51.387753+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.7833333333333333,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-2b-q4km","process_cpu_percent_avg":14.926939453226597,"process_rss_bytes_avg":2006719510.7555552,"process_rss_bytes_peak":2334687232.0,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core","sample_count":182,"sample_interval_ms":249.7748705191043,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6527320064.0},"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T16:36:39.120802+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"cb061291446b8dbd79f524a4fefcd4a4406e2a0d","model_id":"unsloth/qwen3.5-4b","model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-4b-q4km","process_cpu_percent_avg":9.78397589283007,"process_rss_bytes_avg":7517191058.123175,"process_rss_bytes_peak":10744905728.0,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":261,"sample_interval_ms":276.8212083626713,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6344114176.0},"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":4482403488,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T11:31:50.661853+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q8","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","parameters_b":4.0,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","model_key":"qwen3.5-4b-q8","process_cpu_percent_avg":6.023120834777095,"process_rss_bytes_avg":7696510827.330371,"process_rss_bytes_peak":10978181120.0,"run_id":"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core","sample_count":278,"sample_interval_ms":297.275809231578,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":7347224576.0},"run_id":"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T12:38:16.531029+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.9333333333333333,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-9b-q4km","process_cpu_percent_avg":6.3201481230234045,"process_rss_bytes_avg":8711409521.940317,"process_rss_bytes_peak":11538563072.0,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","sample_count":345,"sample_interval_ms":281.76636477524414,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":7018201088.0},"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":4121781472,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T11:47:21.635679+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-ud-q2kxl","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"qwen/qwen3.5-9b-ud-q2kxl","model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","parameters_b":9.0,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"profile":"core","provider_key":"korgis","quantization":"UD-Q2_K_XL","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","model_key":"qwen3.5-9b-ud-q2kxl","process_cpu_percent_avg":4.725490643337322,"process_rss_bytes_avg":7851091423.1669855,"process_rss_bytes_peak":11102388224.0,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core","sample_count":336,"sample_interval_ms":289.6220233097523,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":8726102016.0},"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":6192222208,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-06T14:37:09.228523+00:00","deployment":"local","effective_model_id":"qwen3.8-27b-iq1s","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","experiment_kind":"standard","failure_count":0,"family":"qwen3.8","generation_policy":{},"git_commit":"640138c3429882a42797d28074f98d3fea98a63b","model_id":"Qwen/Qwen3.8-27B","model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","parameters_b":27.0,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8,"profile":"core","provider_key":"korgis","quantization":"UD-IQ1_S","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","model_key":"qwen3.8-27b-iq1s","process_cpu_percent_avg":5.983418081521953,"process_rss_bytes_avg":9550237266.862207,"process_rss_bytes_peak":13159825408.0,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","sample_count":792,"sample_interval_ms":301.02958077118194,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5182652416.0},"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-06T09:01:09.715786+00:00","deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"xhtoken/spark-x2.5-4b","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","parameters_b":null,"practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.95,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"spark-x2.5-4b-q4km","process_cpu_percent_avg":9.07249993238987,"process_rss_bytes_avg":3570442922.666667,"process_rss_bytes_peak":4295933952.0,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core","sample_count":244,"sample_interval_ms":280.7728714734669,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6034112512.0},"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core","runtime_key":"korgis-local","sample_count":60,"task_id":"qa-abstention"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-06T09:35:22.771531+00:00","deployment":"local","effective_model_id":"bonsai-27b-q10","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7833333333333333,"profile":"core","provider_key":"korgis","quantization":"Q1_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"bonsai-27b-q10","process_cpu_percent_avg":9.52518999663954,"process_rss_bytes_avg":6054163284.918202,"process_rss_bytes_peak":8022573056.0,"run_id":"2026-10-core-v1-bonsai-27b-q10-structured-output-core","sample_count":990,"sample_interval_ms":317.9005326878062,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5717721088.0},"run_id":"2026-10-core-v1-bonsai-27b-q10-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T13:59:13.872786+00:00","deployment":"api","effective_model_id":"gemini-3.5-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.5","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.03333333333333333,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core","runtime_key":"gemini-api","sample_count":60,"task_id":"structured-output"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T13:38:42.203597+00:00","deployment":"api","effective_model_id":"gemini-3.8-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","experiment_kind":"standard","failure_count":0,"family":"gemini-3.8","generation_policy":{},"git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","model_id":"gemini-3.8-flash","model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.0,"profile":"core","provider_key":"gemini","quantization":null,"resource_summary":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core","runtime_key":"gemini-api","sample_count":60,"task_id":"structured-output"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-06T12:15:04.761551+00:00","deployment":"api","effective_model_id":"gpt-5.6-luna","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.9166666666666666,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-structured-output-core","runtime_key":"openai-api","sample_count":60,"task_id":"structured-output"},{"artifact_format":null,"artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T13:10:27.524208+00:00","deployment":"api","effective_model_id":"gpt-6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","experiment_kind":"standard","failure_count":0,"family":"gpt-6","generation_policy":{},"git_commit":"7f796c10d99e934da6417ea54405fbf9352ad399","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.8333333333333334,"profile":"core","provider_key":"openai","quantization":null,"resource_summary":null,"run_id":"gpt-6-luna-core-gpt-6-luna-structured-output-core","runtime_key":"openai-api","sample_count":60,"task_id":"structured-output"},{"artifact_format":"gguf","artifact_size_bytes":10947100928,"benchmark_signature":"sha256:consolidated:fd230e4e9f030196ab13d72a0beb260f92fbe3bfb1cfa331ed8e0678b55123f9","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-22T13:12:52.242438+00:00","deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","experiment_kind":"consolidated","failure_count":0,"family":"lfm2.5","generation_policy":{"generation_configs":[{"extra":{},"max_output_tokens":2048,"seed":42,"stop":[],"temperature":0.0},{"extra":{},"max_output_tokens":4096,"seed":42,"stop":[],"temperature":0.0}],"mixed_generation":true},"git_commit":null,"model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7666666666666667,"profile":"core","provider_key":"korgis","quantization":"UD-Q3_K_XL","resource_summary":null,"run_id":"lfm25-consolidated-latest","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T11:56:03.800158+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7333333333333333,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b","process_cpu_percent_avg":8.190733436222855,"process_rss_bytes_avg":4822041698.065609,"process_rss_bytes_peak":6531989504.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-structured-output-core","sample_count":423,"sample_interval_ms":283.9724207012575,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5921701888.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T13:13:01.309902+00:00","deployment":"local","effective_model_id":"nemotron-nano-4b-q8","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7166666666666667,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b-q8","process_cpu_percent_avg":4.8684869265121,"process_rss_bytes_avg":5683850551.952017,"process_rss_bytes_peak":6735822848.0,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core","sample_count":654,"sample_interval_ms":307.429391476698,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5616762880.0},"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T13:35:30.102922+00:00","deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.6833333333333333,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-0.8b-q4km","process_cpu_percent_avg":20.24050747900639,"process_rss_bytes_avg":2703435980.8,"process_rss_bytes_peak":4575264768.0,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core","sample_count":195,"sample_interval_ms":237.40003127216673,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":13519536128.0},"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"},{"artifact_format":"gguf","artifact_size_bytes":3775709216,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T11:21:16.650276+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-bf16","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","parameters_b":2.0,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7166666666666667,"profile":"core","provider_key":"korgis","quantization":"BF16","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","model_key":"qwen3.5-2b-bf16","process_cpu_percent_avg":5.386938078405328,"process_rss_bytes_avg":4179186975.941509,"process_rss_bytes_peak":4437606400.0,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core","sample_count":391,"sample_interval_ms":306.911332594406,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4923883520.0},"run_id":"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T10:55:49.921242+00:00","deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7333333333333333,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-2b-q4km","process_cpu_percent_avg":15.462938913292911,"process_rss_bytes_avg":2241664864.711111,"process_rss_bytes_peak":2811346944.0,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core","sample_count":253,"sample_interval_ms":275.0343119529134,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6283788288.0},"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T16:36:39.120802+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"cb061291446b8dbd79f524a4fefcd4a4406e2a0d","model_id":"unsloth/qwen3.5-4b","model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7333333333333333,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-4b-q4km","process_cpu_percent_avg":8.486342485499302,"process_rss_bytes_avg":5121284030.324117,"process_rss_bytes_peak":6889406464.0,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":411,"sample_interval_ms":286.355363495796,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5873369088.0},"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"},{"artifact_format":"gguf","artifact_size_bytes":4482403488,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T11:30:38.611895+00:00","deployment":"local","effective_model_id":"qwen3.5-4b-q8","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","parameters_b":4.0,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7333333333333333,"profile":"core","provider_key":"korgis","quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","model_key":"qwen3.5-4b-q8","process_cpu_percent_avg":5.698576220486698,"process_rss_bytes_avg":6661104577.102827,"process_rss_bytes_peak":9700769792.0,"run_id":"qwen-variants-core-qwen3.5-4b-q8-structured-output-core","sample_count":477,"sample_interval_ms":317.01940462127334,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5716574208.0},"run_id":"qwen-variants-core-qwen3.5-4b-q8-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T12:36:44.930375+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7666666666666667,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-9b-q4km","process_cpu_percent_avg":5.8239325570441824,"process_rss_bytes_avg":6475378858.969259,"process_rss_bytes_peak":7243612160.0,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core","sample_count":617,"sample_interval_ms":298.72389312365516,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5452595200.0},"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"},{"artifact_format":"gguf","artifact_size_bytes":4121781472,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T11:45:53.885678+00:00","deployment":"local","effective_model_id":"qwen3.5-9b-ud-q2kxl","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","experiment_kind":"standard","failure_count":0,"family":"qwen3.5","generation_policy":{},"git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","model_id":"qwen/qwen3.5-9b-ud-q2kxl","model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","parameters_b":9.0,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7166666666666667,"profile":"core","provider_key":"korgis","quantization":"UD-Q2_K_XL","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","model_key":"qwen3.5-9b-ud-q2kxl","process_cpu_percent_avg":4.974081455251981,"process_rss_bytes_avg":7070526205.54296,"process_rss_bytes_peak":10778755072.0,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core","sample_count":559,"sample_interval_ms":311.91939930876663,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5600329728.0},"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"},{"artifact_format":"gguf","artifact_size_bytes":6192222208,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-06T14:37:06.404488+00:00","deployment":"local","effective_model_id":"qwen3.8-27b-iq1s","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","experiment_kind":"standard","failure_count":0,"family":"qwen3.8","generation_policy":{},"git_commit":"640138c3429882a42797d28074f98d3fea98a63b","model_id":"Qwen/Qwen3.8-27B","model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","parameters_b":27.0,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.6833333333333333,"profile":"core","provider_key":"korgis","quantization":"UD-IQ1_S","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","model_key":"qwen3.8-27b-iq1s","process_cpu_percent_avg":5.429397449534366,"process_rss_bytes_avg":7678658248.931642,"process_rss_bytes_peak":9314566144.0,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","sample_count":1314,"sample_interval_ms":323.2460322738765,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4307910656.0},"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"},{"artifact_format":"gguf","artifact_size_bytes":null,"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-06T09:00:08.668486+00:00","deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","experiment_kind":"standard","failure_count":0,"family":null,"generation_policy":{},"git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","model_id":"xhtoken/spark-x2.5-4b","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","parameters_b":null,"practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7666666666666667,"profile":"core","provider_key":"korgis","quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"spark-x2.5-4b-q4km","process_cpu_percent_avg":7.982890411325502,"process_rss_bytes_avg":3362284766.8271284,"process_rss_bytes_peak":3593027584.0,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core","sample_count":417,"sample_interval_ms":298.61884293800324,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5815320576.0},"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core","runtime_key":"korgis-local","sample_count":60,"task_id":"structured-output"}],"consolidations":[{"capability_id":"intent-classification","generation_policy":{"generation_configs":[{"extra":{},"max_output_tokens":2048,"seed":42,"stop":[],"temperature":0.0},{"extra":{},"max_output_tokens":4096,"seed":42,"stop":[],"temperature":0.0}],"mixed_generation":true},"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-consolidated-latest"},{"capability_id":"mathematical-reasoning","generation_policy":{"generation_configs":[{"extra":{},"max_output_tokens":2048,"seed":42,"stop":[],"temperature":0.0},{"extra":{},"max_output_tokens":4096,"seed":42,"stop":[],"temperature":0.0}],"mixed_generation":true},"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-consolidated-latest"},{"capability_id":"oos-calibration","generation_policy":{"generation_configs":[{"extra":{},"max_output_tokens":2048,"seed":42,"stop":[],"temperature":0.0},{"extra":{},"max_output_tokens":4096,"seed":42,"stop":[],"temperature":0.0}],"mixed_generation":true},"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-consolidated-latest"},{"capability_id":"qa-abstention","generation_policy":{"generation_configs":[{"extra":{},"max_output_tokens":2048,"seed":42,"stop":[],"temperature":0.0},{"extra":{},"max_output_tokens":4096,"seed":42,"stop":[],"temperature":0.0}],"mixed_generation":true},"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-consolidated-latest"},{"capability_id":"structured-output","generation_policy":{"generation_configs":[{"extra":{},"max_output_tokens":2048,"seed":42,"stop":[],"temperature":0.0},{"extra":{},"max_output_tokens":4096,"seed":42,"stop":[],"temperature":0.0}],"mixed_generation":true},"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-consolidated-latest"}],"decision":{"capability_summaries":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":8698.848109240469,"latency_p50_ms":8900.03939550661,"latency_p95_ms":9739.512333347375,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","normalized_quality_score":59.74025974025974,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1529,"primary_metric":"accuracy","primary_value":0.5974025974025974,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-bonsai-27b-q10-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","failure_count":1,"failure_rate":0.006493506493506494,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1155.2659180455662,"latency_p50_ms":1149.7726044981391,"latency_p95_ms":1208.5230768483598,"model_key":"decisio-qwen3.5-2b-q4km","model_signature":"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95","normalized_quality_score":60.3896103896104,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1203,"primary_metric":"accuracy","primary_value":0.6038961038961039,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"1e48c427-4e75-4ace-a465-2aec7e2751cd","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","failure_count":1,"failure_rate":0.006493506493506494,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2834.3784266303433,"latency_p50_ms":2828.9587710023625,"latency_p95_ms":2931.071354152664,"model_key":"decisio-qwen3.5-4b-q4km","model_signature":"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a","normalized_quality_score":70.12987012987013,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1192,"primary_metric":"accuracy","primary_value":0.7012987012987013,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"b906e312-aa49-4397-b60e-a1c3ad6a8f87","sample_count":154},{"benchmark_signature":"sha256:benchmark:718e761e4ba3410c289a5bba91a6474827caaa8da30e644b3df925ba4d16fbfc","capability_id":"intent-classification","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","failure_count":0,"failure_rate":0.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":5000.302291662143,"latency_p50_ms":4986.015040994971,"latency_p95_ms":5194.435066799633,"model_key":"decisio-qwen3.5-9b-q4km","model_signature":"sha256:model:ae0099eeb8925bc828c92072ba2f486986899854d2d3de76e5689cd383c783e3","normalized_quality_score":66.23376623376623,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":603,"primary_metric":"accuracy","primary_value":0.6623376623376623,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"run_id":"e9c4ec37-433f-4b06-b397-9d9b3a4f7480","sample_count":77},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":154,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":131.37987608282052,"latency_p50_ms":127.4633540015202,"latency_p95_ms":156.07859991432633,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","normalized_quality_score":0.0,"observed_case_count":154,"observed_quality_latency_pareto":true,"output_tokens_total":null,"primary_metric":"accuracy","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":154,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":130.65657031785867,"latency_p50_ms":127.60222947690636,"latency_p95_ms":170.61041467823082,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","normalized_quality_score":0.0,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":null,"primary_metric":"accuracy","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.8-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","failure_count":4,"failure_rate":0.025974025974025976,"input_tokens_total":147621,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1730.830770566101,"latency_p50_ms":1359.8202709981706,"latency_p95_ms":4205.3196875946005,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","normalized_quality_score":82.46753246753246,"observed_case_count":154,"observed_quality_latency_pareto":true,"output_tokens_total":4896,"primary_metric":"accuracy","primary_value":0.8246753246753247,"provider_cost_coverage_rate":0.974025974025974,"provider_cost_known":true,"provider_cost_observed_total_usd":0.035399400000000004,"provider_cost_per_1k_cases_usd":0.23599600000000004,"provider_cost_per_case_usd":0.00023599600000000004,"provider_cost_priced_cases":150,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":0,"failure_rate":0.0,"input_tokens_total":151555,"known_provider_cost_quality_pareto":true,"latency_mean_ms":2025.4564632668444,"latency_p50_ms":1870.263104501646,"latency_p95_ms":3181.1156249023043,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","normalized_quality_score":79.87012987012987,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":8558,"primary_metric":"accuracy","primary_value":0.7987012987012987,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0405806,"provider_cost_per_1k_cases_usd":0.26351038961038964,"provider_cost_per_case_usd":0.00026351038961038963,"provider_cost_priced_cases":154,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":154,"provider_cost_total_usd":0.0405806,"run_id":"gpt-6-luna-core-gpt-6-luna-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3","failure_count":0,"failure_rate":0.0,"input_tokens_total":525524,"known_provider_cost_quality_pareto":true,"latency_mean_ms":254.11093069761864,"latency_p50_ms":243.48435449792305,"latency_p95_ms":309.60259374624,"model_key":"jev-1.13.0","model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","normalized_quality_score":74.67532467532467,"observed_case_count":308,"observed_quality_latency_pareto":true,"output_tokens_total":255096,"primary_metric":"accuracy","primary_value":0.7467532467532467,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.022072008,"provider_cost_per_1k_cases_usd":0.07166236363636364,"provider_cost_per_case_usd":7.166236363636363e-05,"provider_cost_priced_cases":308,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"jev","processing":"standard","source":"TypeSafe AI public pricing: $42 per billion input tokens; output free.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":308,"provider_cost_total_usd":0.022072008,"run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:consolidated:42ab94943101512840504e431240a2dc1e9c2c51d6e055498c7b6b44c803f515","capability_id":"intent-classification","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","failure_count":6,"failure_rate":0.03896103896103896,"input_tokens_total":96208,"known_provider_cost_quality_pareto":false,"latency_mean_ms":23087.311111795778,"latency_p50_ms":16176.900271035265,"latency_p95_ms":70185.2069376211,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","normalized_quality_score":70.12987012987013,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":159929,"primary_metric":"accuracy","primary_value":0.7012987012987013,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"lfm25-consolidated-latest","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":97089,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2025.159793870857,"latency_p50_ms":2010.3751250062487,"latency_p95_ms":2163.9121060507023,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","normalized_quality_score":55.84415584415584,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1669,"primary_metric":"accuracy","primary_value":0.5584415584415584,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":97089,"known_provider_cost_quality_pareto":false,"latency_mean_ms":3292.909132578643,"latency_p50_ms":3393.2761040050536,"latency_p95_ms":4267.191872800322,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","normalized_quality_score":55.1948051948052,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1678,"primary_metric":"accuracy","primary_value":0.551948051948052,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":342.47120458998586,"latency_p50_ms":339.46391649806174,"latency_p95_ms":371.62911084524234,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","normalized_quality_score":27.92207792207792,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1492,"primary_metric":"accuracy","primary_value":0.2792207792207792,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":785.6293642306065,"latency_p50_ms":779.2861880152486,"latency_p95_ms":870.3237018693471,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","normalized_quality_score":57.7922077922078,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1509,"primary_metric":"accuracy","primary_value":0.577922077922078,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":699.1592043047497,"latency_p50_ms":688.2556045020465,"latency_p95_ms":780.7678268472954,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","normalized_quality_score":57.7922077922078,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1500,"primary_metric":"accuracy","primary_value":0.577922077922078,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1596.720603324247,"latency_p50_ms":1580.0608540012036,"latency_p95_ms":1735.5302432952158,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","normalized_quality_score":66.88311688311688,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1508,"primary_metric":"accuracy","primary_value":0.6688311688311688,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1682.3662307401423,"latency_p50_ms":1674.4573339819908,"latency_p95_ms":1877.7022937458241,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","normalized_quality_score":64.93506493506493,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1515,"primary_metric":"accuracy","primary_value":0.6493506493506493,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2811.8513665976934,"latency_p50_ms":2785.3132915042806,"latency_p95_ms":3057.8174914451665,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","normalized_quality_score":69.48051948051948,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1507,"primary_metric":"accuracy","primary_value":0.6948051948051948,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2741.8135949013626,"latency_p50_ms":2726.615708990721,"latency_p95_ms":2938.3117294928525,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","normalized_quality_score":68.83116883116884,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1499,"primary_metric":"accuracy","primary_value":0.6883116883116883,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","failure_count":2,"failure_rate":0.012987012987012988,"input_tokens_total":84820,"known_provider_cost_quality_pareto":false,"latency_mean_ms":8791.119202123857,"latency_p50_ms":8568.107791492366,"latency_p95_ms":9098.29579560901,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","normalized_quality_score":58.44155844155844,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1449,"primary_metric":"accuracy","primary_value":0.5844155844155844,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":99556,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2224.7655532798503,"latency_p50_ms":2207.8276665051817,"latency_p95_ms":2593.4791601015604,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","normalized_quality_score":64.93506493506493,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":2139,"primary_metric":"accuracy","primary_value":0.6493506493506493,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:80688f79e2c27746cabb9c8e268a98d63ee3d732cb5c73d70fc581a1958017fc","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2179.467997944448,"latency_p50_ms":2134.854041505605,"latency_p95_ms":2559.561714925802,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","normalized_quality_score":47.5,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":357,"primary_metric":"final_answer_accuracy","primary_value":0.475,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":35,"failure_rate":0.875,"input_tokens_total":429,"known_provider_cost_quality_pareto":false,"latency_mean_ms":433.91745427506976,"latency_p50_ms":175.63410446746275,"latency_p95_ms":2067.160929192323,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","normalized_quality_score":12.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":34,"primary_metric":"final_answer_accuracy","primary_value":0.125,"provider_cost_coverage_rate":0.125,"provider_cost_known":true,"provider_cost_observed_total_usd":4.2375e-05,"provider_cost_per_1k_cases_usd":0.008475,"provider_cost_per_case_usd":8.475e-06,"provider_cost_priced_cases":5,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":0,"failure_rate":0.0,"input_tokens_total":4879,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1708.4744781750487,"latency_p50_ms":1610.1467295084149,"latency_p95_ms":2667.998251819517,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","normalized_quality_score":67.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":1724,"primary_metric":"final_answer_accuracy","primary_value":0.675,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0030446,"provider_cost_per_1k_cases_usd":0.076115,"provider_cost_per_case_usd":7.6115e-05,"provider_cost_priced_cases":40,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":40,"provider_cost_total_usd":0.0030446,"run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","sample_count":40},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":0,"failure_rate":0.0,"input_tokens_total":4879,"known_provider_cost_quality_pareto":true,"latency_mean_ms":1804.218880116241,"latency_p50_ms":1617.0925420010462,"latency_p95_ms":2760.3690872143484,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","normalized_quality_score":70.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":1670,"primary_metric":"final_answer_accuracy","primary_value":0.7,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0029798,"provider_cost_per_1k_cases_usd":0.07449499999999999,"provider_cost_per_case_usd":7.4495e-05,"provider_cost_priced_cases":40,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":40,"provider_cost_total_usd":0.0029798,"run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","sample_count":40},{"benchmark_signature":"sha256:consolidated:20f65d6394555078bc9dce0c1393e10d963aecb84ce9344ece12046a78d42de3","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","failure_count":0,"failure_rate":0.0,"input_tokens_total":4186,"known_provider_cost_quality_pareto":false,"latency_mean_ms":9138.54978744057,"latency_p50_ms":8435.180062428117,"latency_p95_ms":14727.237971138664,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","normalized_quality_score":95.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":20062,"primary_metric":"final_answer_accuracy","primary_value":0.95,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"lfm25-consolidated-latest","sample_count":40},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","failure_count":0,"failure_rate":0.0,"input_tokens_total":2437,"known_provider_cost_quality_pareto":false,"latency_mean_ms":14415.438993828138,"latency_p50_ms":12581.906624487601,"latency_p95_ms":29393.43815765458,"model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","normalized_quality_score":90.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":23911,"primary_metric":"final_answer_accuracy","primary_value":0.9,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"16a842b0-5dcc-47b5-8808-850b2ad1ac10","sample_count":40},{"benchmark_signature":"sha256:benchmark:dca92e0e83499d6053d9585c0bdee4321b7ec9651563312492a659578feb94ee","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:f4c681453ad106c8c878d601ee4a0ee6335674e81abaf3f7ba5297ce35841db0","failure_count":10,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":21810.495604295284,"latency_p50_ms":21736.863624944817,"latency_p95_ms":22921.96444603032,"model_key":"limite-1b-violetto-mlx","model_signature":"sha256:model:fddf01c0845257f40bbb635a92054df8eb22dacd450b827a951962e906d6513b","normalized_quality_score":0.0,"observed_case_count":10,"observed_quality_latency_pareto":false,"output_tokens_total":null,"primary_metric":"final_answer_accuracy","primary_value":0.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":10,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":10,"provider_cost_total_usd":null,"run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","sample_count":10},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:0caf2fff221781464037ba78ff0ed5a8158d632ab0e66481fc8df733dd8785c6","failure_count":0,"failure_rate":0.0,"input_tokens_total":4401,"known_provider_cost_quality_pareto":false,"latency_mean_ms":645.015773922205,"latency_p50_ms":618.5829375172034,"latency_p95_ms":876.0035978746599,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","normalized_quality_score":50.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":325,"primary_metric":"final_answer_accuracy","primary_value":0.5,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"3b0292d4-8f8f-4843-900c-946c32476964","sample_count":40},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:89a74e35df26090fe2c6adb16b01e32373fbcdad6966a1b23cc9b88ba35a5170","failure_count":0,"failure_rate":0.0,"input_tokens_total":4401,"known_provider_cost_quality_pareto":false,"latency_mean_ms":630.3394749935251,"latency_p50_ms":605.8572710026056,"latency_p95_ms":712.8466440772164,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","normalized_quality_score":47.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":335,"primary_metric":"final_answer_accuracy","primary_value":0.475,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:c58f103ecf1090aab4510c6708cd9446fefc4137c61ae9ca33da8cc1c5d3faf8","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":159.1911823285045,"latency_p50_ms":154.29127059178427,"latency_p95_ms":192.66062675160353,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","normalized_quality_score":10.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":345,"primary_metric":"final_answer_accuracy","primary_value":0.1,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":423.17552603199147,"latency_p50_ms":399.95974994963035,"latency_p95_ms":596.0868246445898,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","normalized_quality_score":25.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":351,"primary_metric":"final_answer_accuracy","primary_value":0.25,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:8190c8f3bb1f7b51ef597577c913ceae05f8cb7998a7d5221e848f67071166c9","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":264.1106479219161,"latency_p50_ms":243.21977101499215,"latency_p95_ms":369.4675250735594,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","normalized_quality_score":25.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":343,"primary_metric":"final_answer_accuracy","primary_value":0.25,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"9fb9f932-1f29-4646-b337-f33476d841ba","sample_count":40},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:d6ef7d6b6d33f70d69c5d1df4568941686a610cda0bde994801472a902500dd4","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":550.896484364057,"latency_p50_ms":541.7198544600978,"latency_p95_ms":659.4784934190102,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","normalized_quality_score":40.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":324,"primary_metric":"final_answer_accuracy","primary_value":0.4,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":638.2667771249544,"latency_p50_ms":647.2247504862025,"latency_p95_ms":751.9698580726981,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","normalized_quality_score":47.5,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":325,"primary_metric":"final_answer_accuracy","primary_value":0.475,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:9d728f687c9bb0f47a2c5a8c0e61b6da5a7e2833ad1afcffb18c7b4eb2767886","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":890.66908233508,"latency_p50_ms":886.8744795327075,"latency_p95_ms":1010.3516017552464,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","normalized_quality_score":50.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":327,"primary_metric":"final_answer_accuracy","primary_value":0.5,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:d26122294632fb1ef7e62445eece13d61030562e5a3d7a64b60606c3c12ff3be","failure_count":0,"failure_rate":0.0,"input_tokens_total":4493,"known_provider_cost_quality_pareto":false,"latency_mean_ms":483.35021561942995,"latency_p50_ms":480.80327099887654,"latency_p95_ms":562.0952548051717,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:8dcf43803761fd5570f637e166b655565b6c638e51a6a96848f58f6e6699a734","normalized_quality_score":30.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":346,"primary_metric":"final_answer_accuracy","primary_value":0.3,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"e68754ab-25e6-4449-b06a-66dbc311d9e5","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2219.316747926132,"latency_p50_ms":2213.482687504438,"latency_p95_ms":2470.525904195528,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","normalized_quality_score":57.49999999999999,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":396,"primary_metric":"final_answer_accuracy","primary_value":0.575,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":40,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":134.92541666782927,"latency_p50_ms":129.18631246429868,"latency_p95_ms":178.56887083326,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","normalized_quality_score":0.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":null,"primary_metric":"final_answer_accuracy","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":40,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":135.2158302499447,"latency_p50_ms":129.64654198731296,"latency_p95_ms":177.9962440428789,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","normalized_quality_score":0.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":null,"primary_metric":"final_answer_accuracy","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.8-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4745,"known_provider_cost_quality_pareto":true,"latency_mean_ms":1729.619964624726,"latency_p50_ms":1645.4874375049258,"latency_p95_ms":2669.140464547672,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","normalized_quality_score":45.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":1415,"primary_metric":"final_answer_accuracy","primary_value":0.45,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.002647,"provider_cost_per_1k_cases_usd":0.066175,"provider_cost_per_case_usd":6.6175e-05,"provider_cost_priced_cases":40,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":40,"provider_cost_total_usd":0.002647,"run_id":"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":0,"failure_rate":0.0,"input_tokens_total":4745,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1796.6679405944888,"latency_p50_ms":1630.790083494503,"latency_p95_ms":2689.1042798262743,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","normalized_quality_score":42.5,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":1577,"primary_metric":"final_answer_accuracy","primary_value":0.425,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0028414,"provider_cost_per_1k_cases_usd":0.071035,"provider_cost_per_case_usd":7.1035e-05,"provider_cost_priced_cases":40,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":40,"provider_cost_total_usd":0.0028414,"run_id":"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:600e8b1d94c1e2bc55bd8802918ce0c6f18dece6d8a9c02247a530fc7b015409","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:503d302009e21bb18588b87ac3897f9112994a479905bbca872384efa4360298","failure_count":0,"failure_rate":0.0,"input_tokens_total":4068,"known_provider_cost_quality_pareto":false,"latency_mean_ms":9981.023337302031,"latency_p50_ms":8236.826166510582,"latency_p95_ms":23799.921930843244,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","normalized_quality_score":77.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":22004,"primary_metric":"final_answer_accuracy","primary_value":0.775,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"lfm25-math-v2-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","failure_count":2,"failure_rate":0.05,"input_tokens_total":2323,"known_provider_cost_quality_pareto":false,"latency_mean_ms":16638.17807911255,"latency_p50_ms":12906.520646007266,"latency_p95_ms":49143.83464824058,"model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","normalized_quality_score":80.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":27084,"primary_metric":"final_answer_accuracy","primary_value":0.8,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"8eef4bab-617a-4638-9e51-53e72034e821","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":4295,"known_provider_cost_quality_pareto":false,"latency_mean_ms":574.8850365260296,"latency_p50_ms":566.8227080022916,"latency_p95_ms":664.7058172944526,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","normalized_quality_score":57.49999999999999,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":326,"primary_metric":"final_answer_accuracy","primary_value":0.575,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":4295,"known_provider_cost_quality_pareto":false,"latency_mean_ms":682.1980853499554,"latency_p50_ms":653.759083004843,"latency_p95_ms":789.8509145503339,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","normalized_quality_score":60.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":326,"primary_metric":"final_answer_accuracy","primary_value":0.6,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":162.14768229983747,"latency_p50_ms":147.21993750572437,"latency_p95_ms":202.90230384853203,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","normalized_quality_score":20.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":410,"primary_metric":"final_answer_accuracy","primary_value":0.2,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":402.97180729248794,"latency_p50_ms":390.3209579875693,"latency_p95_ms":501.28230001137115,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","normalized_quality_score":37.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":371,"primary_metric":"final_answer_accuracy","primary_value":0.375,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":266.83754889891134,"latency_p50_ms":261.64224999956787,"latency_p95_ms":332.65334790921764,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","normalized_quality_score":32.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":371,"primary_metric":"final_answer_accuracy","primary_value":0.325,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":557.498613624557,"latency_p50_ms":543.7640624950291,"latency_p95_ms":709.5650934068543,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","normalized_quality_score":47.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":328,"primary_metric":"final_answer_accuracy","primary_value":0.475,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":571.6949749272317,"latency_p50_ms":558.6196244985331,"latency_p95_ms":686.1018520867218,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","normalized_quality_score":50.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":328,"primary_metric":"final_answer_accuracy","primary_value":0.5,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":857.8194458481448,"latency_p50_ms":846.0788744923775,"latency_p95_ms":999.0503857392467,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","normalized_quality_score":67.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":333,"primary_metric":"final_answer_accuracy","primary_value":0.675,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":817.9967738789856,"latency_p50_ms":805.0282710173633,"latency_p95_ms":935.7486851222346,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","normalized_quality_score":55.00000000000001,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":329,"primary_metric":"final_answer_accuracy","primary_value":0.55,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","failure_count":7,"failure_rate":0.175,"input_tokens_total":4075,"known_provider_cost_quality_pareto":false,"latency_mean_ms":5066.640841649496,"latency_p50_ms":2834.1644589963835,"latency_p95_ms":22229.72979789554,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","normalized_quality_score":25.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":719,"primary_metric":"final_answer_accuracy","primary_value":0.25,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":4373,"known_provider_cost_quality_pareto":false,"latency_mean_ms":619.3483061986626,"latency_p50_ms":629.36987500143,"latency_p95_ms":752.714208359248,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","normalized_quality_score":45.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":346,"primary_metric":"final_answer_accuracy","primary_value":0.45,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core","sample_count":40},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":93984,"known_provider_cost_quality_pareto":false,"latency_mean_ms":8717.480386934201,"latency_p50_ms":8677.232062495023,"latency_p95_ms":9162.466885308095,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","normalized_quality_score":94.8051948051948,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":3634,"primary_metric":"oos_detection_rate","primary_value":0.948051948051948,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":154,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":137.23104790891762,"latency_p50_ms":126.3813539990224,"latency_p95_ms":179.44677726482038,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","normalized_quality_score":50.0,"observed_case_count":154,"observed_quality_latency_pareto":true,"output_tokens_total":null,"primary_metric":"oos_detection_rate","primary_value":0.5,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":154,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":130.86163958908034,"latency_p50_ms":128.1952289864421,"latency_p95_ms":157.77898573724084,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","normalized_quality_score":50.0,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":null,"primary_metric":"oos_detection_rate","primary_value":0.5,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.8-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","failure_count":4,"failure_rate":0.025974025974025976,"input_tokens_total":157214,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1541.9887418898584,"latency_p50_ms":1287.3492500002612,"latency_p95_ms":2838.6875687552674,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","normalized_quality_score":100.0,"observed_case_count":154,"observed_quality_latency_pareto":true,"output_tokens_total":5434,"primary_metric":"oos_detection_rate","primary_value":1.0,"provider_cost_coverage_rate":0.974025974025974,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0379636,"provider_cost_per_1k_cases_usd":0.2530906666666667,"provider_cost_per_case_usd":0.00025309066666666667,"provider_cost_priced_cases":150,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":1,"failure_rate":0.006493506493506494,"input_tokens_total":160355,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2620.117901461184,"latency_p50_ms":2474.8345834668726,"latency_p95_ms":4364.541489130352,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","normalized_quality_score":99.35064935064936,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":8856,"primary_metric":"oos_detection_rate","primary_value":0.9935064935064936,"provider_cost_coverage_rate":0.9935064935064936,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0426982,"provider_cost_per_1k_cases_usd":0.27907320261437907,"provider_cost_per_case_usd":0.00027907320261437907,"provider_cost_priced_cases":153,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"gpt-6-luna-core-gpt-6-luna-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3","failure_count":0,"failure_rate":0.0,"input_tokens_total":536778,"known_provider_cost_quality_pareto":true,"latency_mean_ms":262.0940212340925,"latency_p50_ms":256.2655620058649,"latency_p95_ms":318.280074550421,"model_key":"jev-1.13.0","model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","normalized_quality_score":97.40259740259741,"observed_case_count":308,"observed_quality_latency_pareto":true,"output_tokens_total":256742,"primary_metric":"oos_detection_rate","primary_value":0.974025974025974,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.022544676000000003,"provider_cost_per_1k_cases_usd":0.07319700000000001,"provider_cost_per_case_usd":7.319700000000001e-05,"provider_cost_priced_cases":308,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"jev","processing":"standard","source":"TypeSafe AI public pricing: $42 per billion input tokens; output free.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":308,"provider_cost_total_usd":0.022544676000000003,"run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:consolidated:9468d75aebe9e805cefc3643f61a9e5fe6127c7a4e3a987608b30841b9bc9fa0","capability_id":"oos-calibration","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","failure_count":9,"failure_rate":0.05844155844155844,"input_tokens_total":102172,"known_provider_cost_quality_pareto":false,"latency_mean_ms":18534.834868738777,"latency_p50_ms":10469.861020508688,"latency_p95_ms":73937.57543984102,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","normalized_quality_score":95.45454545454545,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":113346,"primary_metric":"oos_detection_rate","primary_value":0.9545454545454546,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"lfm25-consolidated-latest","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":106024,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2420.62463879885,"latency_p50_ms":2395.2548329980345,"latency_p95_ms":2632.3175227080355,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","normalized_quality_score":96.1038961038961,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":3217,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":106024,"known_provider_cost_quality_pareto":false,"latency_mean_ms":3123.8922102271176,"latency_p50_ms":2692.1658124992973,"latency_p95_ms":4934.0581772929,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","normalized_quality_score":96.1038961038961,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":3335,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":93984,"known_provider_cost_quality_pareto":false,"latency_mean_ms":442.98370453869614,"latency_p50_ms":433.2297294968157,"latency_p95_ms":463.41581654851325,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","normalized_quality_score":50.0,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":2864,"primary_metric":"oos_detection_rate","primary_value":0.5,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","failure_count":0,"failure_rate":0.0,"input_tokens_total":93984,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1091.3614762078385,"latency_p50_ms":1085.873708507279,"latency_p95_ms":1192.1733147173654,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","normalized_quality_score":50.0,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":2894,"primary_metric":"oos_detection_rate","primary_value":0.5,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":93984,"known_provider_cost_quality_pareto":false,"latency_mean_ms":817.716475922101,"latency_p50_ms":811.0212289975607,"latency_p95_ms":881.2015722527576,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","normalized_quality_score":50.0,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":2960,"primary_metric":"oos_detection_rate","primary_value":0.5,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":93984,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1858.605324116637,"latency_p50_ms":1845.0444789996254,"latency_p95_ms":1977.098264250526,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","normalized_quality_score":97.40259740259741,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":2637,"primary_metric":"oos_detection_rate","primary_value":0.974025974025974,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","failure_count":0,"failure_rate":0.0,"input_tokens_total":93984,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2090.62336576844,"latency_p50_ms":2040.7832085038535,"latency_p95_ms":2489.353190278052,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","normalized_quality_score":98.05194805194806,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":2663,"primary_metric":"oos_detection_rate","primary_value":0.9805194805194806,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":93984,"known_provider_cost_quality_pareto":false,"latency_mean_ms":5469.462600662698,"latency_p50_ms":5220.9697289945325,"latency_p95_ms":8752.915718645818,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","normalized_quality_score":97.40259740259741,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":2600,"primary_metric":"oos_detection_rate","primary_value":0.974025974025974,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","failure_count":0,"failure_rate":0.0,"input_tokens_total":93984,"known_provider_cost_quality_pareto":false,"latency_mean_ms":3174.727481104624,"latency_p50_ms":3154.345499962801,"latency_p95_ms":3386.294052284211,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","normalized_quality_score":96.1038961038961,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":2629,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","failure_count":16,"failure_rate":0.1038961038961039,"input_tokens_total":84236,"known_provider_cost_quality_pareto":false,"latency_mean_ms":12654.062560350236,"latency_p50_ms":9886.834312499559,"latency_p95_ms":36285.84456239332,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","normalized_quality_score":81.81818181818183,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":2434,"primary_metric":"oos_detection_rate","primary_value":0.8181818181818182,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":108643,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2389.1731853761853,"latency_p50_ms":2335.455271000683,"latency_p95_ms":2798.3671812500684,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","normalized_quality_score":96.1038961038961,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":2950,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","sample_count":154},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":3175.179379200805,"latency_p50_ms":3054.529208500753,"latency_p95_ms":3838.2669604550756,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","normalized_quality_score":86.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1305,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":59,"failure_rate":0.9833333333333333,"input_tokens_total":120,"known_provider_cost_quality_pareto":false,"latency_mean_ms":244.35933958253977,"latency_p50_ms":190.14506248640828,"latency_p95_ms":281.4437542168883,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","normalized_quality_score":1.6666666666666667,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":15,"primary_metric":"qa_correct","primary_value":0.016666666666666666,"provider_cost_coverage_rate":0.016666666666666666,"provider_cost_known":true,"provider_cost_observed_total_usd":1.35e-05,"provider_cost_per_1k_cases_usd":0.0135,"provider_cost_per_case_usd":1.35e-05,"provider_cost_priced_cases":1,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":60,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":136.61597294752332,"latency_p50_ms":129.0290629840456,"latency_p95_ms":191.78036249941215,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","normalized_quality_score":0.0,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":null,"primary_metric":"qa_correct","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.8-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","failure_count":0,"failure_rate":0.0,"input_tokens_total":9763,"known_provider_cost_quality_pareto":true,"latency_mean_ms":1337.4706270170766,"latency_p50_ms":1291.5158120013075,"latency_p95_ms":1930.584752104187,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","normalized_quality_score":86.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1370,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0035966,"provider_cost_per_1k_cases_usd":0.059943333333333335,"provider_cost_per_case_usd":5.994333333333333e-05,"provider_cost_priced_cases":60,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":60,"provider_cost_total_usd":0.0035966,"run_id":"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":0,"failure_rate":0.0,"input_tokens_total":9763,"known_provider_cost_quality_pareto":true,"latency_mean_ms":1647.2096979327034,"latency_p50_ms":1527.1968334855046,"latency_p95_ms":2859.7770288557504,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","normalized_quality_score":95.0,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1869,"primary_metric":"qa_correct","primary_value":0.95,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0041954,"provider_cost_per_1k_cases_usd":0.06992333333333334,"provider_cost_per_case_usd":6.992333333333334e-05,"provider_cost_priced_cases":60,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":60,"provider_cost_total_usd":0.0041954,"run_id":"gpt-6-luna-core-gpt-6-luna-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:consolidated:b832fe612fcfe138845fe193a557c686f68ecb2f59d1b3f4784e653e9ec82e5c","capability_id":"qa-abstention","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","failure_count":0,"failure_rate":0.0,"input_tokens_total":8170,"known_provider_cost_quality_pareto":false,"latency_mean_ms":8713.557179173222,"latency_p50_ms":7030.449937446974,"latency_p95_ms":18365.44755841605,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","normalized_quality_score":96.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":23009,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"lfm25-consolidated-latest","sample_count":60},{"benchmark_signature":"sha256:benchmark:1fa41ef05579230896c951f4e01da57ccf719647ad6f8235008243556409a7cb","capability_id":"qa-abstention","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","failure_count":12,"failure_rate":1.0,"input_tokens_total":803,"known_provider_cost_quality_pareto":false,"latency_mean_ms":44512.825486056194,"latency_p50_ms":50686.88316695625,"latency_p95_ms":53967.651789076626,"model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","normalized_quality_score":0.0,"observed_case_count":12,"observed_quality_latency_pareto":false,"output_tokens_total":20683,"primary_metric":"qa_correct","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":12,"provider_cost_total_usd":null,"run_id":"b6b4123a-1d20-4c39-8644-ea00c1e5893b","sample_count":12},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":8421,"known_provider_cost_quality_pareto":false,"latency_mean_ms":860.4363417003091,"latency_p50_ms":849.5050004930818,"latency_p95_ms":1040.7841770924276,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","normalized_quality_score":96.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":897,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":8421,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1412.304399300774,"latency_p50_ms":1362.8550000066753,"latency_p95_ms":1983.9075827017948,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","normalized_quality_score":96.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":877,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":212.94066396609801,"latency_p50_ms":204.67102049587993,"latency_p95_ms":269.198137501371,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","normalized_quality_score":48.333333333333336,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":999,"primary_metric":"qa_correct","primary_value":0.48333333333333334,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","failure_count":10,"failure_rate":0.16666666666666666,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":665.2421306236647,"latency_p50_ms":616.8598124640994,"latency_p95_ms":1006.9639482157072,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","normalized_quality_score":63.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1014,"primary_metric":"qa_correct","primary_value":0.6333333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":423.106404900318,"latency_p50_ms":398.915562494949,"latency_p95_ms":549.0888670014099,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","normalized_quality_score":78.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":1025,"primary_metric":"qa_correct","primary_value":0.7833333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":876.0588624500087,"latency_p50_ms":786.4938955026446,"latency_p95_ms":1081.4400829891365,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","normalized_quality_score":86.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1085,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":989.199661828267,"latency_p50_ms":908.3277915196959,"latency_p95_ms":1313.4596493822753,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","normalized_quality_score":86.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1036,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1270.930086817437,"latency_p50_ms":1220.2731874931487,"latency_p95_ms":1647.5607910033425,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","normalized_quality_score":93.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":930,"primary_metric":"qa_correct","primary_value":0.9333333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1255.4060783363336,"latency_p50_ms":1200.648166501196,"latency_p95_ms":1755.204953884822,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","normalized_quality_score":86.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1014,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","failure_count":1,"failure_rate":0.016666666666666666,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":3615.463377833172,"latency_p50_ms":3467.9861039985553,"latency_p95_ms":4320.6283913576035,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","normalized_quality_score":80.0,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":961,"primary_metric":"qa_correct","primary_value":0.8,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"failure_rate":0.0,"input_tokens_total":8552,"known_provider_cost_quality_pareto":false,"latency_mean_ms":772.222803482388,"latency_p50_ms":746.6509164951276,"latency_p95_ms":985.1788809428398,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","normalized_quality_score":95.0,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":871,"primary_metric":"qa_correct","primary_value":0.95,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":6,"failure_rate":0.1,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":4839.719958333929,"latency_p50_ms":4484.091166501457,"latency_p95_ms":7536.193280648876,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","normalized_quality_score":78.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2689,"primary_metric":"exact_match","primary_value":0.7833333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-bonsai-27b-q10-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":58,"failure_rate":0.9666666666666667,"input_tokens_total":242,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1515.5912111426005,"latency_p50_ms":160.70060449419543,"latency_p95_ms":2834.6031663066206,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","normalized_quality_score":3.3333333333333335,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":64,"primary_metric":"exact_match","primary_value":0.03333333333333333,"provider_cost_coverage_rate":0.03333333333333333,"provider_cost_known":true,"provider_cost_observed_total_usd":3.735e-05,"provider_cost_per_1k_cases_usd":0.018675,"provider_cost_per_case_usd":1.8675e-05,"provider_cost_priced_cases":2,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":60,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":4986.970465994091,"latency_p50_ms":192.07681200350635,"latency_p95_ms":60002.98497467884,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","normalized_quality_score":0.0,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":null,"primary_metric":"exact_match","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.8-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","failure_count":0,"failure_rate":0.0,"input_tokens_total":10570,"known_provider_cost_quality_pareto":true,"latency_mean_ms":1934.9246188348236,"latency_p50_ms":1730.0426670044544,"latency_p95_ms":2594.3603166953817,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","normalized_quality_score":91.66666666666666,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":3316,"primary_metric":"exact_match","primary_value":0.9166666666666666,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0060932,"provider_cost_per_1k_cases_usd":0.10155333333333333,"provider_cost_per_case_usd":0.00010155333333333333,"provider_cost_priced_cases":60,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":60,"provider_cost_total_usd":0.0060932,"run_id":"2026-10-core-v1-gpt-5.6-luna-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":0,"failure_rate":0.0,"input_tokens_total":10570,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2258.3810763889537,"latency_p50_ms":2046.710938011529,"latency_p95_ms":3548.6162735644007,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","normalized_quality_score":83.33333333333334,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":3356,"primary_metric":"exact_match","primary_value":0.8333333333333334,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0061412,"provider_cost_per_1k_cases_usd":0.10235333333333334,"provider_cost_per_case_usd":0.00010235333333333333,"provider_cost_priced_cases":60,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":60,"provider_cost_total_usd":0.0061412,"run_id":"gpt-6-luna-core-gpt-6-luna-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:consolidated:fd230e4e9f030196ab13d72a0beb260f92fbe3bfb1cfa331ed8e0678b55123f9","capability_id":"structured-output","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","failure_count":0,"failure_rate":0.0,"input_tokens_total":8212,"known_provider_cost_quality_pareto":false,"latency_mean_ms":11260.75794856879,"latency_p50_ms":9417.586916533764,"latency_p95_ms":22641.866754152565,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","normalized_quality_score":76.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":35103,"primary_metric":"exact_match","primary_value":0.7666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"lfm25-consolidated-latest","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":2,"failure_rate":0.03333333333333333,"input_tokens_total":8424,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1666.822604850798,"latency_p50_ms":1567.1223544995883,"latency_p95_ms":2798.8641288560757,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","normalized_quality_score":73.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2345,"primary_metric":"exact_match","primary_value":0.7333333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":2,"failure_rate":0.03333333333333333,"input_tokens_total":8424,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2980.6488194495614,"latency_p50_ms":2663.501458497194,"latency_p95_ms":5424.1780169439135,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","normalized_quality_score":71.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2346,"primary_metric":"exact_match","primary_value":0.7166666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":3,"failure_rate":0.05,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":473.98608951771166,"latency_p50_ms":432.87633300496964,"latency_p95_ms":798.6456500082567,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","normalized_quality_score":68.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":2661,"primary_metric":"exact_match","primary_value":0.6833333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","failure_count":5,"failure_rate":0.08333333333333333,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1611.7719881663409,"latency_p50_ms":1479.928854474565,"latency_p95_ms":3047.5117707916065,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","normalized_quality_score":71.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2624,"primary_metric":"exact_match","primary_value":0.7166666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":4,"failure_rate":0.06666666666666667,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":806.0470264669372,"latency_p50_ms":733.9355000003707,"latency_p95_ms":1450.6030038486642,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","normalized_quality_score":73.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":2621,"primary_metric":"exact_match","primary_value":0.7333333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":3,"failure_rate":0.05,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1623.2633048988646,"latency_p50_ms":1483.7232709978707,"latency_p95_ms":2773.320496148517,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","normalized_quality_score":73.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2638,"primary_metric":"exact_match","primary_value":0.7333333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","failure_count":2,"failure_rate":0.03333333333333333,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2109.738486816059,"latency_p50_ms":1936.254354484845,"latency_p95_ms":3802.9144431435266,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","normalized_quality_score":73.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2621,"primary_metric":"exact_match","primary_value":0.7333333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-4b-q8-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":2,"failure_rate":0.03333333333333333,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2716.7342707665134,"latency_p50_ms":2522.184229499544,"latency_p95_ms":4643.15792289417,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","normalized_quality_score":76.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2658,"primary_metric":"exact_match","primary_value":0.7666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","failure_count":4,"failure_rate":0.06666666666666667,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2517.9370430715303,"latency_p50_ms":2332.108625007095,"latency_p95_ms":4228.940910479287,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","normalized_quality_score":71.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2696,"primary_metric":"exact_match","primary_value":0.7166666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","failure_count":10,"failure_rate":0.16666666666666666,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":6660.125209748609,"latency_p50_ms":5688.188729494868,"latency_p95_ms":12916.947212500963,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","normalized_quality_score":68.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2395,"primary_metric":"exact_match","primary_value":0.6833333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","sample_count":60},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":1,"failure_rate":0.016666666666666666,"input_tokens_total":8652,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1693.6658090327303,"latency_p50_ms":1434.1119370001252,"latency_p95_ms":3149.145677108754,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","normalized_quality_score":76.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":2690,"primary_metric":"exact_match","primary_value":0.7666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core","sample_count":60}],"cohort_policy":{"benchmark_signatures":{"intent-classification":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","mathematical-reasoning":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","mathematical-reasoning-v2":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","oos-calibration":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","qa-abstention":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","structured-output":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1"},"description":"Per capability, use the CURRENT benchmark lineage with the largest model coverage; break ties by latest completion.","policy_id":"max-model-coverage-latest-v1"},"dataset_summaries":[{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":8698.848109240469,"latency_p50_ms":8900.03939550661,"latency_p95_ms":9739.512333347375,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","normalized_quality_score":59.74025974025974,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1529,"primary_metric":"accuracy","primary_value":0.5974025974025974,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2179.467997944448,"latency_p50_ms":2134.854041505605,"latency_p95_ms":2559.561714925802,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","normalized_quality_score":47.5,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":357,"primary_metric":"final_answer_accuracy","primary_value":0.475,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2219.316747926132,"latency_p50_ms":2213.482687504438,"latency_p95_ms":2470.525904195528,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","normalized_quality_score":57.49999999999999,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":396,"primary_metric":"final_answer_accuracy","primary_value":0.575,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":46999,"known_provider_cost_quality_pareto":false,"latency_mean_ms":8778.879600687424,"latency_p50_ms":8735.466374986572,"latency_p95_ms":9121.223125007236,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","normalized_quality_score":98.7012987012987,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1924,"primary_metric":"oos_correct","primary_value":0.987012987012987,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":46985,"known_provider_cost_quality_pareto":false,"latency_mean_ms":8656.081173180975,"latency_p50_ms":8601.950458993088,"latency_p95_ms":9165.10655801103,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","normalized_quality_score":90.9090909090909,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1710,"primary_metric":"oos_correct","primary_value":0.9090909090909091,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":3175.179379200805,"latency_p50_ms":3054.529208500753,"latency_p95_ms":3838.2669604550756,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","normalized_quality_score":86.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1305,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":6,"failure_rate":0.1,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":4839.719958333929,"latency_p50_ms":4484.091166501457,"latency_p95_ms":7536.193280648876,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","normalized_quality_score":78.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2689,"primary_metric":"exact_match","primary_value":0.7833333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":1,"failure_rate":0.006493506493506494,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1155.2659180455662,"latency_p50_ms":1149.7726044981391,"latency_p95_ms":1208.5230768483598,"model_key":"decisio-qwen3.5-2b-q4km","model_signature":"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95","normalized_quality_score":60.3896103896104,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1203,"primary_metric":"accuracy","primary_value":0.6038961038961039,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":1,"failure_rate":0.006493506493506494,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2834.3784266303433,"latency_p50_ms":2828.9587710023625,"latency_p95_ms":2931.071354152664,"model_key":"decisio-qwen3.5-4b-q4km","model_signature":"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a","normalized_quality_score":70.12987012987013,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1192,"primary_metric":"accuracy","primary_value":0.7012987012987013,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":5000.302291662143,"latency_p50_ms":4986.015040994971,"latency_p95_ms":5194.435066799633,"model_key":"decisio-qwen3.5-9b-q4km","model_signature":"sha256:model:ae0099eeb8925bc828c92072ba2f486986899854d2d3de76e5689cd383c783e3","normalized_quality_score":66.23376623376623,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":603,"primary_metric":"accuracy","primary_value":0.6623376623376623,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":154,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":131.37987608282052,"latency_p50_ms":127.4633540015202,"latency_p95_ms":156.07859991432633,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","normalized_quality_score":0.0,"observed_case_count":154,"observed_quality_latency_pareto":true,"output_tokens_total":null,"primary_metric":"accuracy","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":35,"failure_rate":0.875,"input_tokens_total":429,"known_provider_cost_quality_pareto":false,"latency_mean_ms":433.91745427506976,"latency_p50_ms":175.63410446746275,"latency_p95_ms":2067.160929192323,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","normalized_quality_score":12.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":34,"primary_metric":"final_answer_accuracy","primary_value":0.125,"provider_cost_coverage_rate":0.125,"provider_cost_known":true,"provider_cost_observed_total_usd":4.2375e-05,"provider_cost_per_1k_cases_usd":0.008475,"provider_cost_per_case_usd":8.475e-06,"provider_cost_priced_cases":5,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":40,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":134.92541666782927,"latency_p50_ms":129.18631246429868,"latency_p95_ms":178.56887083326,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","normalized_quality_score":0.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":null,"primary_metric":"final_answer_accuracy","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":77,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":136.1645048986153,"latency_p50_ms":126.8837500247173,"latency_p95_ms":177.19717521686113,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","normalized_quality_score":0.0,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":null,"primary_metric":"oos_correct","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":77,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":138.29759091921997,"latency_p50_ms":123.73316701268777,"latency_p95_ms":178.8445335580036,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","normalized_quality_score":0.0,"observed_case_count":77,"observed_quality_latency_pareto":true,"output_tokens_total":null,"primary_metric":"oos_correct","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":59,"failure_rate":0.9833333333333333,"input_tokens_total":120,"known_provider_cost_quality_pareto":false,"latency_mean_ms":244.35933958253977,"latency_p50_ms":190.14506248640828,"latency_p95_ms":281.4437542168883,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","normalized_quality_score":1.6666666666666667,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":15,"primary_metric":"qa_correct","primary_value":0.016666666666666666,"provider_cost_coverage_rate":0.016666666666666666,"provider_cost_known":true,"provider_cost_observed_total_usd":1.35e-05,"provider_cost_per_1k_cases_usd":0.0135,"provider_cost_per_case_usd":1.35e-05,"provider_cost_priced_cases":1,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":58,"failure_rate":0.9666666666666667,"input_tokens_total":242,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1515.5912111426005,"latency_p50_ms":160.70060449419543,"latency_p95_ms":2834.6031663066206,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","normalized_quality_score":3.3333333333333335,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":64,"primary_metric":"exact_match","primary_value":0.03333333333333333,"provider_cost_coverage_rate":0.03333333333333333,"provider_cost_known":true,"provider_cost_observed_total_usd":3.735e-05,"provider_cost_per_1k_cases_usd":0.018675,"provider_cost_per_case_usd":1.8675e-05,"provider_cost_priced_cases":2,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":154,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":130.65657031785867,"latency_p50_ms":127.60222947690636,"latency_p95_ms":170.61041467823082,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","normalized_quality_score":0.0,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":null,"primary_metric":"accuracy","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.8-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":40,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":135.2158302499447,"latency_p50_ms":129.64654198731296,"latency_p95_ms":177.9962440428789,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","normalized_quality_score":0.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":null,"primary_metric":"final_answer_accuracy","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.8-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":77,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":130.14289070005444,"latency_p50_ms":126.64366699755192,"latency_p95_ms":157.0738750160672,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","normalized_quality_score":0.0,"observed_case_count":77,"observed_quality_latency_pareto":true,"output_tokens_total":null,"primary_metric":"oos_correct","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.8-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":77,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":131.58038847810624,"latency_p50_ms":129.40495798829943,"latency_p95_ms":156.79174198303375,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","normalized_quality_score":0.0,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":null,"primary_metric":"oos_correct","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.8-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":60,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":136.61597294752332,"latency_p50_ms":129.0290629840456,"latency_p95_ms":191.78036249941215,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","normalized_quality_score":0.0,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":null,"primary_metric":"qa_correct","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.8-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":60,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":4986.970465994091,"latency_p50_ms":192.07681200350635,"latency_p95_ms":60002.98497467884,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","normalized_quality_score":0.0,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":null,"primary_metric":"exact_match","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.8-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":4,"failure_rate":0.025974025974025976,"input_tokens_total":147621,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1730.830770566101,"latency_p50_ms":1359.8202709981706,"latency_p95_ms":4205.3196875946005,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","normalized_quality_score":82.46753246753246,"observed_case_count":154,"observed_quality_latency_pareto":true,"output_tokens_total":4896,"primary_metric":"accuracy","primary_value":0.8246753246753247,"provider_cost_coverage_rate":0.974025974025974,"provider_cost_known":true,"provider_cost_observed_total_usd":0.035399400000000004,"provider_cost_per_1k_cases_usd":0.23599600000000004,"provider_cost_per_case_usd":0.00023599600000000004,"provider_cost_priced_cases":150,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4879,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1708.4744781750487,"latency_p50_ms":1610.1467295084149,"latency_p95_ms":2667.998251819517,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","normalized_quality_score":67.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":1724,"primary_metric":"final_answer_accuracy","primary_value":0.675,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0030446,"provider_cost_per_1k_cases_usd":0.076115,"provider_cost_per_case_usd":7.6115e-05,"provider_cost_priced_cases":40,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":40,"provider_cost_total_usd":0.0030446,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4745,"known_provider_cost_quality_pareto":true,"latency_mean_ms":1729.619964624726,"latency_p50_ms":1645.4874375049258,"latency_p95_ms":2669.140464547672,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","normalized_quality_score":45.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":1415,"primary_metric":"final_answer_accuracy","primary_value":0.45,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.002647,"provider_cost_per_1k_cases_usd":0.066175,"provider_cost_per_case_usd":6.6175e-05,"provider_cost_priced_cases":40,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":40,"provider_cost_total_usd":0.002647,"sample_count":40},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":4,"failure_rate":0.05194805194805195,"input_tokens_total":76515,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1893.9702813515466,"latency_p50_ms":1548.235541995382,"latency_p95_ms":3869.6360582049238,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","normalized_quality_score":94.8051948051948,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":3750,"primary_metric":"oos_correct","primary_value":0.948051948051948,"provider_cost_coverage_rate":0.948051948051948,"provider_cost_known":true,"provider_cost_observed_total_usd":0.019803,"provider_cost_per_1k_cases_usd":0.2712739726027398,"provider_cost_per_case_usd":0.00027127397260273976,"provider_cost_priced_cases":73,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":80699,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1190.0072024281699,"latency_p50_ms":1147.1069999970496,"latency_p95_ms":1587.2184588079108,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","normalized_quality_score":100.0,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1684,"primary_metric":"oos_correct","primary_value":1.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.018160600000000002,"provider_cost_per_1k_cases_usd":0.2358519480519481,"provider_cost_per_case_usd":0.0002358519480519481,"provider_cost_priced_cases":77,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":77,"provider_cost_total_usd":0.018160600000000002,"sample_count":77},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":9763,"known_provider_cost_quality_pareto":true,"latency_mean_ms":1337.4706270170766,"latency_p50_ms":1291.5158120013075,"latency_p95_ms":1930.584752104187,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","normalized_quality_score":86.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1370,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0035966,"provider_cost_per_1k_cases_usd":0.059943333333333335,"provider_cost_per_case_usd":5.994333333333333e-05,"provider_cost_priced_cases":60,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":60,"provider_cost_total_usd":0.0035966,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":10570,"known_provider_cost_quality_pareto":true,"latency_mean_ms":1934.9246188348236,"latency_p50_ms":1730.0426670044544,"latency_p95_ms":2594.3603166953817,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","normalized_quality_score":91.66666666666666,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":3316,"primary_metric":"exact_match","primary_value":0.9166666666666666,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0060932,"provider_cost_per_1k_cases_usd":0.10155333333333333,"provider_cost_per_case_usd":0.00010155333333333333,"provider_cost_priced_cases":60,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":60,"provider_cost_total_usd":0.0060932,"sample_count":60},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":151555,"known_provider_cost_quality_pareto":true,"latency_mean_ms":2025.4564632668444,"latency_p50_ms":1870.263104501646,"latency_p95_ms":3181.1156249023043,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","normalized_quality_score":79.87012987012987,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":8558,"primary_metric":"accuracy","primary_value":0.7987012987012987,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0405806,"provider_cost_per_1k_cases_usd":0.26351038961038964,"provider_cost_per_case_usd":0.00026351038961038963,"provider_cost_priced_cases":154,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":154,"provider_cost_total_usd":0.0405806,"sample_count":154},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4879,"known_provider_cost_quality_pareto":true,"latency_mean_ms":1804.218880116241,"latency_p50_ms":1617.0925420010462,"latency_p95_ms":2760.3690872143484,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","normalized_quality_score":70.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":1670,"primary_metric":"final_answer_accuracy","primary_value":0.7,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0029798,"provider_cost_per_1k_cases_usd":0.07449499999999999,"provider_cost_per_case_usd":7.4495e-05,"provider_cost_priced_cases":40,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":40,"provider_cost_total_usd":0.0029798,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4745,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1796.6679405944888,"latency_p50_ms":1630.790083494503,"latency_p95_ms":2689.1042798262743,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","normalized_quality_score":42.5,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":1577,"primary_metric":"final_answer_accuracy","primary_value":0.425,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0028414,"provider_cost_per_1k_cases_usd":0.071035,"provider_cost_per_case_usd":7.1035e-05,"provider_cost_priced_cases":40,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":40,"provider_cost_total_usd":0.0028414,"sample_count":40},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":1,"failure_rate":0.012987012987012988,"input_tokens_total":79656,"known_provider_cost_quality_pareto":false,"latency_mean_ms":3057.293945286187,"latency_p50_ms":2784.830249962397,"latency_p95_ms":4620.6940750125805,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","normalized_quality_score":97.40259740259741,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":5863,"primary_metric":"oos_correct","primary_value":0.974025974025974,"provider_cost_coverage_rate":0.987012987012987,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0229668,"provider_cost_per_1k_cases_usd":0.3021947368421053,"provider_cost_per_case_usd":0.0003021947368421053,"provider_cost_priced_cases":76,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":80699,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2182.9418576361804,"latency_p50_ms":1999.0120829897933,"latency_p95_ms":3477.565124980174,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","normalized_quality_score":100.0,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":2993,"primary_metric":"oos_correct","primary_value":1.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0197314,"provider_cost_per_1k_cases_usd":0.25625194805194806,"provider_cost_per_case_usd":0.0002562519480519481,"provider_cost_priced_cases":77,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":77,"provider_cost_total_usd":0.0197314,"sample_count":77},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":9763,"known_provider_cost_quality_pareto":true,"latency_mean_ms":1647.2096979327034,"latency_p50_ms":1527.1968334855046,"latency_p95_ms":2859.7770288557504,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","normalized_quality_score":95.0,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1869,"primary_metric":"qa_correct","primary_value":0.95,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0041954,"provider_cost_per_1k_cases_usd":0.06992333333333334,"provider_cost_per_case_usd":6.992333333333334e-05,"provider_cost_priced_cases":60,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":60,"provider_cost_total_usd":0.0041954,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":10570,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2258.3810763889537,"latency_p50_ms":2046.710938011529,"latency_p95_ms":3548.6162735644007,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","normalized_quality_score":83.33333333333334,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":3356,"primary_metric":"exact_match","primary_value":0.8333333333333334,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0061412,"provider_cost_per_1k_cases_usd":0.10235333333333334,"provider_cost_per_case_usd":0.00010235333333333333,"provider_cost_priced_cases":60,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":60,"provider_cost_total_usd":0.0061412,"sample_count":60},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":525524,"known_provider_cost_quality_pareto":true,"latency_mean_ms":254.11093069761864,"latency_p50_ms":243.48435449792305,"latency_p95_ms":309.60259374624,"model_key":"jev-1.13.0","model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","normalized_quality_score":74.67532467532467,"observed_case_count":308,"observed_quality_latency_pareto":true,"output_tokens_total":255096,"primary_metric":"accuracy","primary_value":0.7467532467532467,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.022072008,"provider_cost_per_1k_cases_usd":0.07166236363636364,"provider_cost_per_case_usd":7.166236363636363e-05,"provider_cost_priced_cases":308,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"jev","processing":"standard","source":"TypeSafe AI public pricing: $42 per billion input tokens; output free.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":308,"provider_cost_total_usd":0.022072008,"sample_count":308},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":268460,"known_provider_cost_quality_pareto":true,"latency_mean_ms":261.2221466434829,"latency_p50_ms":256.74289550079266,"latency_p95_ms":320.20307664934074,"model_key":"jev-1.13.0","model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","normalized_quality_score":94.8051948051948,"observed_case_count":154,"observed_quality_latency_pareto":true,"output_tokens_total":128614,"primary_metric":"oos_correct","primary_value":0.948051948051948,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.01127532,"provider_cost_per_1k_cases_usd":0.07321636363636363,"provider_cost_per_case_usd":7.321636363636364e-05,"provider_cost_priced_cases":154,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"jev","processing":"standard","source":"TypeSafe AI public pricing: $42 per billion input tokens; output free.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":154,"provider_cost_total_usd":0.01127532,"sample_count":154},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":268318,"known_provider_cost_quality_pareto":true,"latency_mean_ms":262.96589582470205,"latency_p50_ms":256.0998745029792,"latency_p95_ms":316.27981425554026,"model_key":"jev-1.13.0","model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","normalized_quality_score":100.0,"observed_case_count":154,"observed_quality_latency_pareto":true,"output_tokens_total":128128,"primary_metric":"oos_correct","primary_value":1.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.011269356000000001,"provider_cost_per_1k_cases_usd":0.07317763636363636,"provider_cost_per_case_usd":7.317763636363637e-05,"provider_cost_priced_cases":154,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"jev","processing":"standard","source":"TypeSafe AI public pricing: $42 per billion input tokens; output free.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":154,"provider_cost_total_usd":0.011269356000000001,"sample_count":154},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":6,"failure_rate":0.03896103896103896,"input_tokens_total":96208,"known_provider_cost_quality_pareto":false,"latency_mean_ms":23087.311111795778,"latency_p50_ms":16176.900271035265,"latency_p95_ms":70185.2069376211,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","normalized_quality_score":70.12987012987013,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":159929,"primary_metric":"accuracy","primary_value":0.7012987012987013,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4186,"known_provider_cost_quality_pareto":false,"latency_mean_ms":9138.54978744057,"latency_p50_ms":8435.180062428117,"latency_p95_ms":14727.237971138664,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","normalized_quality_score":95.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":20062,"primary_metric":"final_answer_accuracy","primary_value":0.95,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4068,"known_provider_cost_quality_pareto":false,"latency_mean_ms":9981.023337302031,"latency_p50_ms":8236.826166510582,"latency_p95_ms":23799.921930843244,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","normalized_quality_score":77.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":22004,"primary_metric":"final_answer_accuracy","primary_value":0.775,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":8,"failure_rate":0.1038961038961039,"input_tokens_total":48636,"known_provider_cost_quality_pareto":false,"latency_mean_ms":25299.56137446094,"latency_p50_ms":12379.735084017739,"latency_p95_ms":74805.01954995561,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","normalized_quality_score":85.71428571428571,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":70948,"primary_metric":"oos_correct","primary_value":0.8571428571428571,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":1,"failure_rate":0.012987012987012988,"input_tokens_total":53536,"known_provider_cost_quality_pareto":false,"latency_mean_ms":11770.108363016618,"latency_p50_ms":9689.080916927196,"latency_p95_ms":24273.661758471288,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","normalized_quality_score":94.8051948051948,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":42398,"primary_metric":"oos_correct","primary_value":0.948051948051948,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":8170,"known_provider_cost_quality_pareto":false,"latency_mean_ms":8713.557179173222,"latency_p50_ms":7030.449937446974,"latency_p95_ms":18365.44755841605,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","normalized_quality_score":96.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":23009,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":8212,"known_provider_cost_quality_pareto":false,"latency_mean_ms":11260.75794856879,"latency_p50_ms":9417.586916533764,"latency_p95_ms":22641.866754152565,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","normalized_quality_score":76.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":35103,"primary_metric":"exact_match","primary_value":0.7666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":2437,"known_provider_cost_quality_pareto":false,"latency_mean_ms":14415.438993828138,"latency_p50_ms":12581.906624487601,"latency_p95_ms":29393.43815765458,"model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","normalized_quality_score":90.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":23911,"primary_metric":"final_answer_accuracy","primary_value":0.9,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":2,"failure_rate":0.05,"input_tokens_total":2323,"known_provider_cost_quality_pareto":false,"latency_mean_ms":16638.17807911255,"latency_p50_ms":12906.520646007266,"latency_p95_ms":49143.83464824058,"model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","normalized_quality_score":80.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":27084,"primary_metric":"final_answer_accuracy","primary_value":0.8,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":12,"failure_rate":1.0,"input_tokens_total":803,"known_provider_cost_quality_pareto":false,"latency_mean_ms":44512.825486056194,"latency_p50_ms":50686.88316695625,"latency_p95_ms":53967.651789076626,"model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","normalized_quality_score":0.0,"observed_case_count":12,"observed_quality_latency_pareto":false,"output_tokens_total":20683,"primary_metric":"qa_correct","primary_value":0.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":12,"provider_cost_total_usd":null,"sample_count":12},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":10,"failure_rate":1.0,"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":21810.495604295284,"latency_p50_ms":21736.863624944817,"latency_p95_ms":22921.96444603032,"model_key":"limite-1b-violetto-mlx","model_signature":"sha256:model:fddf01c0845257f40bbb635a92054df8eb22dacd450b827a951962e906d6513b","normalized_quality_score":0.0,"observed_case_count":10,"observed_quality_latency_pareto":false,"output_tokens_total":null,"primary_metric":"final_answer_accuracy","primary_value":0.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":10,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":10,"provider_cost_total_usd":null,"sample_count":10},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":97089,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2025.159793870857,"latency_p50_ms":2010.3751250062487,"latency_p95_ms":2163.9121060507023,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","normalized_quality_score":55.84415584415584,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1669,"primary_metric":"accuracy","primary_value":0.5584415584415584,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4401,"known_provider_cost_quality_pareto":false,"latency_mean_ms":645.015773922205,"latency_p50_ms":618.5829375172034,"latency_p95_ms":876.0035978746599,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","normalized_quality_score":50.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":325,"primary_metric":"final_answer_accuracy","primary_value":0.5,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4295,"known_provider_cost_quality_pareto":false,"latency_mean_ms":574.8850365260296,"latency_p50_ms":566.8227080022916,"latency_p95_ms":664.7058172944526,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","normalized_quality_score":57.49999999999999,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":326,"primary_metric":"final_answer_accuracy","primary_value":0.575,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":53013,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2499.965928078216,"latency_p50_ms":2489.686166998581,"latency_p95_ms":2695.2199669991387,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","normalized_quality_score":98.7012987012987,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1799,"primary_metric":"oos_correct","primary_value":0.987012987012987,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":53011,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2341.2833495194836,"latency_p50_ms":2314.8674159892835,"latency_p95_ms":2477.2517746081576,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","normalized_quality_score":93.5064935064935,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1418,"primary_metric":"oos_correct","primary_value":0.935064935064935,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":8421,"known_provider_cost_quality_pareto":false,"latency_mean_ms":860.4363417003091,"latency_p50_ms":849.5050004930818,"latency_p95_ms":1040.7841770924276,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","normalized_quality_score":96.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":897,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":2,"failure_rate":0.03333333333333333,"input_tokens_total":8424,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1666.822604850798,"latency_p50_ms":1567.1223544995883,"latency_p95_ms":2798.8641288560757,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","normalized_quality_score":73.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2345,"primary_metric":"exact_match","primary_value":0.7333333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":97089,"known_provider_cost_quality_pareto":false,"latency_mean_ms":3292.909132578643,"latency_p50_ms":3393.2761040050536,"latency_p95_ms":4267.191872800322,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","normalized_quality_score":55.1948051948052,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1678,"primary_metric":"accuracy","primary_value":0.551948051948052,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4401,"known_provider_cost_quality_pareto":false,"latency_mean_ms":630.3394749935251,"latency_p50_ms":605.8572710026056,"latency_p95_ms":712.8466440772164,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","normalized_quality_score":47.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":335,"primary_metric":"final_answer_accuracy","primary_value":0.475,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4295,"known_provider_cost_quality_pareto":false,"latency_mean_ms":682.1980853499554,"latency_p50_ms":653.759083004843,"latency_p95_ms":789.8509145503339,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","normalized_quality_score":60.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":326,"primary_metric":"final_answer_accuracy","primary_value":0.6,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":53013,"known_provider_cost_quality_pareto":false,"latency_mean_ms":3766.431912324541,"latency_p50_ms":3453.931582989753,"latency_p95_ms":5078.716925196933,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","normalized_quality_score":98.7012987012987,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1832,"primary_metric":"oos_correct","primary_value":0.987012987012987,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":53011,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2481.3525081296943,"latency_p50_ms":2494.065542006865,"latency_p95_ms":2691.221599999699,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","normalized_quality_score":93.5064935064935,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1503,"primary_metric":"oos_correct","primary_value":0.935064935064935,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":8421,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1412.304399300774,"latency_p50_ms":1362.8550000066753,"latency_p95_ms":1983.9075827017948,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","normalized_quality_score":96.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":877,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":2,"failure_rate":0.03333333333333333,"input_tokens_total":8424,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2980.6488194495614,"latency_p50_ms":2663.501458497194,"latency_p95_ms":5424.1780169439135,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","normalized_quality_score":71.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2346,"primary_metric":"exact_match","primary_value":0.7166666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":342.47120458998586,"latency_p50_ms":339.46391649806174,"latency_p95_ms":371.62911084524234,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","normalized_quality_score":27.92207792207792,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1492,"primary_metric":"accuracy","primary_value":0.2792207792207792,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":159.1911823285045,"latency_p50_ms":154.29127059178427,"latency_p95_ms":192.66062675160353,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","normalized_quality_score":10.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":345,"primary_metric":"final_answer_accuracy","primary_value":0.1,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":162.14768229983747,"latency_p50_ms":147.21993750572437,"latency_p95_ms":202.90230384853203,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","normalized_quality_score":20.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":410,"primary_metric":"final_answer_accuracy","primary_value":0.2,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":46999,"known_provider_cost_quality_pareto":false,"latency_mean_ms":431.5476136492995,"latency_p50_ms":428.76354200416245,"latency_p95_ms":454.68750780564733,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","normalized_quality_score":100.0,"observed_case_count":77,"observed_quality_latency_pareto":true,"output_tokens_total":1448,"primary_metric":"oos_correct","primary_value":1.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":46985,"known_provider_cost_quality_pareto":false,"latency_mean_ms":454.4197954280928,"latency_p50_ms":437.74258300254587,"latency_p95_ms":470.6647918006638,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","normalized_quality_score":0.0,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1416,"primary_metric":"oos_correct","primary_value":0.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":212.94066396609801,"latency_p50_ms":204.67102049587993,"latency_p95_ms":269.198137501371,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","normalized_quality_score":48.333333333333336,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":999,"primary_metric":"qa_correct","primary_value":0.48333333333333334,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":3,"failure_rate":0.05,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":473.98608951771166,"latency_p50_ms":432.87633300496964,"latency_p95_ms":798.6456500082567,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","normalized_quality_score":68.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":2661,"primary_metric":"exact_match","primary_value":0.6833333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":785.6293642306065,"latency_p50_ms":779.2861880152486,"latency_p95_ms":870.3237018693471,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","normalized_quality_score":57.7922077922078,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1509,"primary_metric":"accuracy","primary_value":0.577922077922078,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":423.17552603199147,"latency_p50_ms":399.95974994963035,"latency_p95_ms":596.0868246445898,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","normalized_quality_score":25.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":351,"primary_metric":"final_answer_accuracy","primary_value":0.25,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":402.97180729248794,"latency_p50_ms":390.3209579875693,"latency_p95_ms":501.28230001137115,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","normalized_quality_score":37.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":371,"primary_metric":"final_answer_accuracy","primary_value":0.375,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":46999,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1085.9827381821785,"latency_p50_ms":1085.135250003077,"latency_p95_ms":1179.1477915947328,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","normalized_quality_score":100.0,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1451,"primary_metric":"oos_correct","primary_value":1.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":46985,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1096.7402142334984,"latency_p50_ms":1086.6620419546962,"latency_p95_ms":1191.4266084204428,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","normalized_quality_score":0.0,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1443,"primary_metric":"oos_correct","primary_value":0.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":10,"failure_rate":0.16666666666666666,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":665.2421306236647,"latency_p50_ms":616.8598124640994,"latency_p95_ms":1006.9639482157072,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","normalized_quality_score":63.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1014,"primary_metric":"qa_correct","primary_value":0.6333333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":5,"failure_rate":0.08333333333333333,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1611.7719881663409,"latency_p50_ms":1479.928854474565,"latency_p95_ms":3047.5117707916065,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","normalized_quality_score":71.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2624,"primary_metric":"exact_match","primary_value":0.7166666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":699.1592043047497,"latency_p50_ms":688.2556045020465,"latency_p95_ms":780.7678268472954,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","normalized_quality_score":57.7922077922078,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1500,"primary_metric":"accuracy","primary_value":0.577922077922078,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":264.1106479219161,"latency_p50_ms":243.21977101499215,"latency_p95_ms":369.4675250735594,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","normalized_quality_score":25.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":343,"primary_metric":"final_answer_accuracy","primary_value":0.25,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":266.83754889891134,"latency_p50_ms":261.64224999956787,"latency_p95_ms":332.65334790921764,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","normalized_quality_score":32.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":371,"primary_metric":"final_answer_accuracy","primary_value":0.325,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":46999,"known_provider_cost_quality_pareto":false,"latency_mean_ms":831.4678630655156,"latency_p50_ms":826.020000007702,"latency_p95_ms":921.0994412045693,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","normalized_quality_score":100.0,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1503,"primary_metric":"oos_correct","primary_value":1.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":46985,"known_provider_cost_quality_pareto":false,"latency_mean_ms":803.9650887786864,"latency_p50_ms":799.419791001128,"latency_p95_ms":844.2399912077235,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","normalized_quality_score":0.0,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1457,"primary_metric":"oos_correct","primary_value":0.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":423.106404900318,"latency_p50_ms":398.915562494949,"latency_p95_ms":549.0888670014099,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","normalized_quality_score":78.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":1025,"primary_metric":"qa_correct","primary_value":0.7833333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":4,"failure_rate":0.06666666666666667,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":806.0470264669372,"latency_p50_ms":733.9355000003707,"latency_p95_ms":1450.6030038486642,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","normalized_quality_score":73.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":2621,"primary_metric":"exact_match","primary_value":0.7333333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1596.720603324247,"latency_p50_ms":1580.0608540012036,"latency_p95_ms":1735.5302432952158,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","normalized_quality_score":66.88311688311688,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1508,"primary_metric":"accuracy","primary_value":0.6688311688311688,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":550.896484364057,"latency_p50_ms":541.7198544600978,"latency_p95_ms":659.4784934190102,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","normalized_quality_score":40.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":324,"primary_metric":"final_answer_accuracy","primary_value":0.4,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":557.498613624557,"latency_p50_ms":543.7640624950291,"latency_p95_ms":709.5650934068543,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","normalized_quality_score":47.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":328,"primary_metric":"final_answer_accuracy","primary_value":0.475,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":46999,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1897.1279539875065,"latency_p50_ms":1879.2896250088234,"latency_p95_ms":2049.794733597082,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","normalized_quality_score":98.7012987012987,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1447,"primary_metric":"oos_correct","primary_value":0.987012987012987,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":46985,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1820.082694245767,"latency_p50_ms":1807.692207992659,"latency_p95_ms":1940.8644089999143,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","normalized_quality_score":96.1038961038961,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1190,"primary_metric":"oos_correct","primary_value":0.961038961038961,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":876.0588624500087,"latency_p50_ms":786.4938955026446,"latency_p95_ms":1081.4400829891365,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","normalized_quality_score":86.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1085,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":3,"failure_rate":0.05,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1623.2633048988646,"latency_p50_ms":1483.7232709978707,"latency_p95_ms":2773.320496148517,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","normalized_quality_score":73.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2638,"primary_metric":"exact_match","primary_value":0.7333333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1682.3662307401423,"latency_p50_ms":1674.4573339819908,"latency_p95_ms":1877.7022937458241,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","normalized_quality_score":64.93506493506493,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1515,"primary_metric":"accuracy","primary_value":0.6493506493506493,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":638.2667771249544,"latency_p50_ms":647.2247504862025,"latency_p95_ms":751.9698580726981,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","normalized_quality_score":47.5,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":325,"primary_metric":"final_answer_accuracy","primary_value":0.475,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":571.6949749272317,"latency_p50_ms":558.6196244985331,"latency_p95_ms":686.1018520867218,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","normalized_quality_score":50.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":328,"primary_metric":"final_answer_accuracy","primary_value":0.5,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":46999,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2208.720919327635,"latency_p50_ms":2165.3660840238445,"latency_p95_ms":2628.3127831760794,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","normalized_quality_score":98.7012987012987,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1488,"primary_metric":"oos_correct","primary_value":0.987012987012987,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":46985,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1972.5258122092453,"latency_p50_ms":1946.7772909556516,"latency_p95_ms":2136.2961751641706,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","normalized_quality_score":97.40259740259741,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1175,"primary_metric":"oos_correct","primary_value":0.974025974025974,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":989.199661828267,"latency_p50_ms":908.3277915196959,"latency_p95_ms":1313.4596493822753,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","normalized_quality_score":86.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1036,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":2,"failure_rate":0.03333333333333333,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2109.738486816059,"latency_p50_ms":1936.254354484845,"latency_p95_ms":3802.9144431435266,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","normalized_quality_score":73.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2621,"primary_metric":"exact_match","primary_value":0.7333333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2811.8513665976934,"latency_p50_ms":2785.3132915042806,"latency_p95_ms":3057.8174914451665,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","normalized_quality_score":69.48051948051948,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1507,"primary_metric":"accuracy","primary_value":0.6948051948051948,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4398,"known_provider_cost_quality_pareto":false,"latency_mean_ms":890.66908233508,"latency_p50_ms":886.8744795327075,"latency_p95_ms":1010.3516017552464,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","normalized_quality_score":50.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":327,"primary_metric":"final_answer_accuracy","primary_value":0.5,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":857.8194458481448,"latency_p50_ms":846.0788744923775,"latency_p95_ms":999.0503857392467,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","normalized_quality_score":67.5,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":333,"primary_metric":"final_answer_accuracy","primary_value":0.675,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":46999,"known_provider_cost_quality_pareto":false,"latency_mean_ms":4758.679551974259,"latency_p50_ms":3863.904249999905,"latency_p95_ms":7737.850358401193,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","normalized_quality_score":97.40259740259741,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1436,"primary_metric":"oos_correct","primary_value":0.974025974025974,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":46985,"known_provider_cost_quality_pareto":false,"latency_mean_ms":6180.245649351136,"latency_p50_ms":6215.419833009946,"latency_p95_ms":8815.064166203956,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","normalized_quality_score":97.40259740259741,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1164,"primary_metric":"oos_correct","primary_value":0.974025974025974,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1270.930086817437,"latency_p50_ms":1220.2731874931487,"latency_p95_ms":1647.5607910033425,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","normalized_quality_score":93.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":930,"primary_metric":"qa_correct","primary_value":0.9333333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":2,"failure_rate":0.03333333333333333,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2716.7342707665134,"latency_p50_ms":2522.184229499544,"latency_p95_ms":4643.15792289417,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","normalized_quality_score":76.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2658,"primary_metric":"exact_match","primary_value":0.7666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":85369,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2741.8135949013626,"latency_p50_ms":2726.615708990721,"latency_p95_ms":2938.3117294928525,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","normalized_quality_score":68.83116883116884,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1499,"primary_metric":"accuracy","primary_value":0.6883116883116883,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4283,"known_provider_cost_quality_pareto":false,"latency_mean_ms":817.9967738789856,"latency_p50_ms":805.0282710173633,"latency_p95_ms":935.7486851222346,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","normalized_quality_score":55.00000000000001,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":329,"primary_metric":"final_answer_accuracy","primary_value":0.55,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":46999,"known_provider_cost_quality_pareto":false,"latency_mean_ms":3249.9961693923883,"latency_p50_ms":3231.6983329947107,"latency_p95_ms":3474.731591984164,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","normalized_quality_score":98.7012987012987,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1435,"primary_metric":"oos_correct","primary_value":0.987012987012987,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":46985,"known_provider_cost_quality_pareto":false,"latency_mean_ms":3099.45879281686,"latency_p50_ms":3090.7840419677086,"latency_p95_ms":3210.958441812545,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","normalized_quality_score":93.5064935064935,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1194,"primary_metric":"oos_correct","primary_value":0.935064935064935,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1255.4060783363336,"latency_p50_ms":1200.648166501196,"latency_p95_ms":1755.204953884822,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","normalized_quality_score":86.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":1014,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":4,"failure_rate":0.06666666666666667,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2517.9370430715303,"latency_p50_ms":2332.108625007095,"latency_p95_ms":4228.940910479287,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","normalized_quality_score":71.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2696,"primary_metric":"exact_match","primary_value":0.7166666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":2,"failure_rate":0.012987012987012988,"input_tokens_total":84820,"known_provider_cost_quality_pareto":false,"latency_mean_ms":8791.119202123857,"latency_p50_ms":8568.107791492366,"latency_p95_ms":9098.29579560901,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","normalized_quality_score":58.44155844155844,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":1449,"primary_metric":"accuracy","primary_value":0.5844155844155844,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":7,"failure_rate":0.175,"input_tokens_total":4075,"known_provider_cost_quality_pareto":false,"latency_mean_ms":5066.640841649496,"latency_p50_ms":2834.1644589963835,"latency_p95_ms":22229.72979789554,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","normalized_quality_score":25.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":719,"primary_metric":"final_answer_accuracy","primary_value":0.25,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":3,"failure_rate":0.03896103896103896,"input_tokens_total":45176,"known_provider_cost_quality_pareto":false,"latency_mean_ms":11085.036545479934,"latency_p50_ms":10003.22662499093,"latency_p95_ms":10811.810099994182,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","normalized_quality_score":93.5064935064935,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1399,"primary_metric":"oos_correct","primary_value":0.935064935064935,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":13,"failure_rate":0.16883116883116883,"input_tokens_total":39060,"known_provider_cost_quality_pareto":false,"latency_mean_ms":14223.088575220538,"latency_p50_ms":9674.696875008522,"latency_p95_ms":36420.40486720216,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","normalized_quality_score":66.23376623376623,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1035,"primary_metric":"oos_correct","primary_value":0.6623376623376623,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":1,"failure_rate":0.016666666666666666,"input_tokens_total":8408,"known_provider_cost_quality_pareto":false,"latency_mean_ms":3615.463377833172,"latency_p50_ms":3467.9861039985553,"latency_p95_ms":4320.6283913576035,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","normalized_quality_score":80.0,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":961,"primary_metric":"qa_correct","primary_value":0.8,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":10,"failure_rate":0.16666666666666666,"input_tokens_total":8346,"known_provider_cost_quality_pareto":false,"latency_mean_ms":6660.125209748609,"latency_p50_ms":5688.188729494868,"latency_p95_ms":12916.947212500963,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","normalized_quality_score":68.33333333333333,"observed_case_count":60,"observed_quality_latency_pareto":false,"output_tokens_total":2395,"primary_metric":"exact_match","primary_value":0.6833333333333333,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"intent-classification","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":99556,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2224.7655532798503,"latency_p50_ms":2207.8276665051817,"latency_p95_ms":2593.4791601015604,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","normalized_quality_score":64.93506493506493,"observed_case_count":154,"observed_quality_latency_pareto":false,"output_tokens_total":2139,"primary_metric":"accuracy","primary_value":0.6493506493506493,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"sample_count":154},{"capability_id":"mathematical-reasoning","dataset_id":"math-reasoning-controlled-v3","failure_count":0,"failure_rate":0.0,"input_tokens_total":4493,"known_provider_cost_quality_pareto":false,"latency_mean_ms":483.35021561942995,"latency_p50_ms":480.80327099887654,"latency_p95_ms":562.0952548051717,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:8dcf43803761fd5570f637e166b655565b6c638e51a6a96848f58f6e6699a734","normalized_quality_score":30.0,"observed_case_count":40,"observed_quality_latency_pareto":true,"output_tokens_total":346,"primary_metric":"final_answer_accuracy","primary_value":0.3,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"mathematical-reasoning-v2","dataset_id":"math-reasoning-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":4373,"known_provider_cost_quality_pareto":false,"latency_mean_ms":619.3483061986626,"latency_p50_ms":629.36987500143,"latency_p95_ms":752.714208359248,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","normalized_quality_score":45.0,"observed_case_count":40,"observed_quality_latency_pareto":false,"output_tokens_total":346,"primary_metric":"final_answer_accuracy","primary_value":0.45,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":40,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":40,"provider_cost_total_usd":null,"sample_count":40},{"capability_id":"oos-calibration","dataset_id":"banking77","failure_count":0,"failure_rate":0.0,"input_tokens_total":54325,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2492.3314697267156,"latency_p50_ms":2434.8514579905896,"latency_p95_ms":2920.9799582022247,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","normalized_quality_score":96.1038961038961,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1631,"primary_metric":"oos_correct","primary_value":0.961038961038961,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"oos-calibration","dataset_id":"clinc150-oos","failure_count":0,"failure_rate":0.0,"input_tokens_total":54318,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2286.0149010256546,"latency_p50_ms":2229.7938340052497,"latency_p95_ms":2551.516925197211,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","normalized_quality_score":96.1038961038961,"observed_case_count":77,"observed_quality_latency_pareto":false,"output_tokens_total":1319,"primary_metric":"oos_correct","primary_value":0.961038961038961,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"sample_count":77},{"capability_id":"qa-abstention","dataset_id":"qa-abstention-controlled-v2","failure_count":0,"failure_rate":0.0,"input_tokens_total":8552,"known_provider_cost_quality_pareto":false,"latency_mean_ms":772.222803482388,"latency_p50_ms":746.6509164951276,"latency_p95_ms":985.1788809428398,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","normalized_quality_score":95.0,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":871,"primary_metric":"qa_correct","primary_value":0.95,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60},{"capability_id":"structured-output","dataset_id":"structured-output-controlled-v2","failure_count":1,"failure_rate":0.016666666666666666,"input_tokens_total":8652,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1693.6658090327303,"latency_p50_ms":1434.1119370001252,"latency_p95_ms":3149.145677108754,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","normalized_quality_score":76.66666666666667,"observed_case_count":60,"observed_quality_latency_pareto":true,"output_tokens_total":2690,"primary_metric":"exact_match","primary_value":0.7666666666666667,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":60,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":60,"provider_cost_total_usd":null,"sample_count":60}],"model_summaries":[{"artifact_format":"gguf","artifact_size_bytes":null,"current_run_ids":["2026-10-core-v1-bonsai-27b-q10-intent-classification-core","2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","2026-10-core-v1-bonsai-27b-q10-oos-calibration-core","2026-10-core-v1-bonsai-27b-q10-qa-abstention-core","2026-10-core-v1-bonsai-27b-q10-structured-output-core","b10ad834-2bdd-409b-9073-49f0233792ea"],"deployment":"local","effective_model_id":"bonsai-27b-q10","execution_count":6,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:80688f79e2c27746cabb9c8e268a98d63ee3d732cb5c73d70fc581a1958017fc","machine":"arm64","release":"25.6.0","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":6,"failure_rate":0.011811023622047244,"family":"bonsai","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":204788,"known_provider_cost_quality_pareto":false,"latency_mean_ms":6572.755784444501,"latency_p50_ms":8354.460334005125,"latency_p95_ms":9424.563083646353,"latest_completed_at_utc":"2026-10-16T08:53:37.155498+00:00","model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","observed_case_count":508,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":9910,"overall_quality_score":70.75757575757575,"parameters_b":27,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":508,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"Q1_0","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","machine":"arm64","release":"25.6.0","run_id":"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","peak_run_id":"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","process_cpu_percent_avg":7.963578373856078,"process_rss_bytes_avg":6587746767.267236,"process_rss_bytes_peak":11685068800.0,"resource_cell_count":6,"sample_count":11302,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4949098496.0},"result_policy":"standard","runtime_key":"korgis-local","tags":["local","open-weight","bonsai","27b"]},{"artifact_format":"gguf","artifact_size_bytes":null,"current_run_ids":["1e48c427-4e75-4ace-a465-2aec7e2751cd"],"deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_count":1,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","machine":"arm64","release":"25.6.0","run_id":"1e48c427-4e75-4ace-a465-2aec7e2751cd","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":1,"failure_rate":0.006493506493506494,"family":"qwen3.5","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1155.2659180455662,"latency_p50_ms":1149.7726044981391,"latency_p95_ms":1208.5230768483598,"latest_completed_at_utc":"2026-10-05T15:26:12.447493+00:00","model_id":"unsloth/qwen3.5-2b","model_key":"decisio-qwen3.5-2b-q4km","model_signature":"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95","observed_case_count":154,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":true,"output_tokens_total":1203,"overall_quality_score":60.3896103896104,"parameters_b":2,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"provider_key":"decisio","quality_coverage_capabilities":1,"quality_coverage_complete":false,"quality_coverage_datasets":1,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":null,"peak_execution_signature":null,"peak_run_id":null,"process_cpu_percent_avg":null,"process_rss_bytes_avg":null,"process_rss_bytes_peak":null,"resource_cell_count":0,"sample_count":0,"sampling_error_count":0,"scope":"unavailable","source":"unavailable","system_available_memory_bytes_min":null},"result_policy":"standard","runtime_key":"decisio-local","tags":["local","decisio","open-weight","qwen","decision-model"]},{"artifact_format":"gguf","artifact_size_bytes":null,"current_run_ids":["b906e312-aa49-4397-b60e-a1c3ad6a8f87"],"deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_count":1,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","machine":"arm64","release":"25.6.0","run_id":"b906e312-aa49-4397-b60e-a1c3ad6a8f87","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":1,"failure_rate":0.006493506493506494,"family":"qwen3.5","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2834.3784266303433,"latency_p50_ms":2828.9587710023625,"latency_p95_ms":2931.071354152664,"latest_completed_at_utc":"2026-10-05T15:33:33.988827+00:00","model_id":"unsloth/qwen3.5-4b","model_key":"decisio-qwen3.5-4b-q4km","model_signature":"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a","observed_case_count":154,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":1192,"overall_quality_score":70.12987012987013,"parameters_b":4,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":154,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":154,"provider_cost_total_usd":null,"provider_key":"decisio","quality_coverage_capabilities":1,"quality_coverage_complete":false,"quality_coverage_datasets":1,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":null,"peak_execution_signature":null,"peak_run_id":null,"process_cpu_percent_avg":null,"process_rss_bytes_avg":null,"process_rss_bytes_peak":null,"resource_cell_count":0,"sample_count":0,"sampling_error_count":0,"scope":"unavailable","source":"unavailable","system_available_memory_bytes_min":null},"result_policy":"standard","runtime_key":"decisio-local","tags":["local","decisio","open-weight","qwen","decision-model"]},{"artifact_format":"gguf","artifact_size_bytes":null,"current_run_ids":["e9c4ec37-433f-4b06-b397-9d9b3a4f7480"],"deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_count":1,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","machine":"arm64","release":"25.6.0","run_id":"e9c4ec37-433f-4b06-b397-9d9b3a4f7480","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"smoke","failure_count":0,"failure_rate":0.0,"family":"qwen3.5","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":5000.302291662143,"latency_p50_ms":4986.015040994971,"latency_p95_ms":5194.435066799633,"latest_completed_at_utc":"2026-10-05T16:17:35.400024+00:00","model_id":"qwen/qwen3.5-9b","model_key":"decisio-qwen3.5-9b-q4km","model_signature":"sha256:model:ae0099eeb8925bc828c92072ba2f486986899854d2d3de76e5689cd383c783e3","observed_case_count":77,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":603,"overall_quality_score":66.23376623376623,"parameters_b":9,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":77,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":77,"provider_cost_total_usd":null,"provider_key":"decisio","quality_coverage_capabilities":1,"quality_coverage_complete":false,"quality_coverage_datasets":1,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":null,"peak_execution_signature":null,"peak_run_id":null,"process_cpu_percent_avg":null,"process_rss_bytes_avg":null,"process_rss_bytes_peak":null,"resource_cell_count":0,"sample_count":0,"sampling_error_count":0,"scope":"unavailable","source":"unavailable","system_available_memory_bytes_min":null},"result_policy":"standard","runtime_key":"decisio-local","tags":["local","decisio","open-weight","qwen","decision-model"]},{"artifact_format":null,"artifact_size_bytes":null,"current_run_ids":["b10ad834-2bdd-409b-9073-49f0233792ea","gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core","gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core","gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core","gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core","gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core"],"deployment":"api","effective_model_id":"gemini-3.5-flash","execution_count":6,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","machine":"arm64","release":"25.6.0","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":500,"failure_rate":0.984251968503937,"family":"gemini-3.5","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":791,"known_provider_cost_quality_pareto":false,"latency_mean_ms":334.08824837785835,"latency_p50_ms":133.2621249894146,"latency_p95_ms":213.5791892651468,"latest_completed_at_utc":"2026-10-16T08:53:37.155498+00:00","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","observed_case_count":508,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":true,"observed_quality_parameters_pareto":false,"output_tokens_total":113,"overall_quality_score":11.25,"parameters_b":null,"provider_cost_coverage_rate":0.015748031496062992,"provider_cost_known":true,"provider_cost_observed_total_usd":9.3225e-05,"provider_cost_per_1k_cases_usd":0.011653125,"provider_cost_per_case_usd":1.1653125e-05,"provider_cost_priced_cases":8,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.5-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"gemini","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":null,"resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":null,"peak_execution_signature":null,"peak_run_id":null,"process_cpu_percent_avg":null,"process_rss_bytes_avg":null,"process_rss_bytes_peak":null,"resource_cell_count":0,"sample_count":0,"sampling_error_count":0,"scope":"unavailable","source":"unavailable","system_available_memory_bytes_min":null},"result_policy":"standard","runtime_key":"gemini-api","tags":["api","closed","google","gemini","flash"]},{"artifact_format":null,"artifact_size_bytes":null,"current_run_ids":["gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core","gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core","gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core","gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core","gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core"],"deployment":"api","effective_model_id":"gemini-3.8-flash","execution_count":5,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","machine":"arm64","release":"25.6.0","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":468,"failure_rate":1.0,"family":"gemini-3.8","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":754.4821022909471,"latency_p50_ms":129.50879096752033,"latency_p95_ms":208.09100856131408,"latest_completed_at_utc":"2026-10-13T13:39:43.452981+00:00","model_id":"gemini-3.8-flash","model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","observed_case_count":468,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":true,"observed_quality_parameters_pareto":false,"output_tokens_total":null,"overall_quality_score":10.0,"parameters_b":null,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gemini-3.8-flash","processing":"standard","source":"Google Gemini API standard pricing snapshot for Flash tier.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"unavailable","provider_cost_total_cases":468,"provider_cost_total_usd":null,"provider_key":"gemini","quality_coverage_capabilities":5,"quality_coverage_complete":false,"quality_coverage_datasets":5,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":null,"resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":null,"peak_execution_signature":null,"peak_run_id":null,"process_cpu_percent_avg":null,"process_rss_bytes_avg":null,"process_rss_bytes_peak":null,"resource_cell_count":0,"sample_count":0,"sampling_error_count":0,"scope":"unavailable","source":"unavailable","system_available_memory_bytes_min":null},"result_policy":"standard","runtime_key":"gemini-api","tags":["api","closed","google","gemini","flash"]},{"artifact_format":null,"artifact_size_bytes":null,"current_run_ids":["2026-10-core-v1-gpt-5.6-luna-intent-classification-core","2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core","2026-10-core-v1-gpt-5.6-luna-oos-calibration-core","2026-10-core-v1-gpt-5.6-luna-qa-abstention-core","2026-10-core-v1-gpt-5.6-luna-structured-output-core","f56a81bc-d911-4b76-9b68-361b14e92a2c"],"deployment":"api","effective_model_id":"gpt-5.6-luna","execution_count":6,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","machine":"arm64","release":"25.6.0","run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":8,"failure_rate":0.015748031496062992,"family":"gpt-5.6","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":334792,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1649.373420041974,"latency_p50_ms":1445.532062505663,"latency_p95_ms":2903.054808448358,"latest_completed_at_utc":"2026-10-15T15:34:31.093143+00:00","model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","observed_case_count":508,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":18155,"overall_quality_score":78.88347763347763,"parameters_b":null,"provider_cost_coverage_rate":0.984251968503937,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0887444,"provider_cost_per_1k_cases_usd":0.1774888,"provider_cost_per_case_usd":0.0001774888,"provider_cost_priced_cases":500,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-5.6-luna","processing":"standard","source":"OpenAI API standard short-context pricing snapshot inherited from the repository Jev benchmark.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"openai","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":null,"resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":null,"peak_execution_signature":null,"peak_run_id":null,"process_cpu_percent_avg":null,"process_rss_bytes_avg":null,"process_rss_bytes_peak":null,"resource_cell_count":0,"sample_count":0,"sampling_error_count":0,"scope":"unavailable","source":"unavailable","system_available_memory_bytes_min":null},"result_policy":"standard","runtime_key":"openai-api","tags":["api","closed","openai"]},{"artifact_format":null,"artifact_size_bytes":null,"current_run_ids":["f56a81bc-d911-4b76-9b68-361b14e92a2c","gpt-6-luna-core-gpt-6-luna-intent-classification-core","gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core","gpt-6-luna-core-gpt-6-luna-oos-calibration-core","gpt-6-luna-core-gpt-6-luna-qa-abstention-core","gpt-6-luna-core-gpt-6-luna-structured-output-core"],"deployment":"api","effective_model_id":"gpt-6-luna","execution_count":6,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","machine":"arm64","release":"25.6.0","run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":1,"failure_rate":0.001968503937007874,"family":"gpt-6","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":341867,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2153.128683968199,"latency_p50_ms":1896.0361664940137,"latency_p95_ms":3750.8784476871365,"latest_completed_at_utc":"2026-10-15T15:34:31.093143+00:00","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","observed_case_count":508,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":25886,"overall_quality_score":78.3423520923521,"parameters_b":null,"provider_cost_coverage_rate":0.9980314960629921,"provider_cost_known":true,"provider_cost_observed_total_usd":0.0994366,"provider_cost_per_1k_cases_usd":0.19612741617357002,"provider_cost_per_case_usd":0.00019612741617357,"provider_cost_priced_cases":507,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"gpt-6-luna","processing":"standard","source":"OpenAI API standard pricing snapshot for Luna family.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"partial","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"openai","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":null,"resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":null,"peak_execution_signature":null,"peak_run_id":null,"process_cpu_percent_avg":null,"process_rss_bytes_avg":null,"process_rss_bytes_peak":null,"resource_cell_count":0,"sample_count":0,"sampling_error_count":0,"scope":"unavailable","source":"unavailable","system_available_memory_bytes_min":null},"result_policy":"standard","runtime_key":"openai-api","tags":["api","closed","openai"]},{"artifact_format":null,"artifact_size_bytes":null,"current_run_ids":["2026-10-core-v1-jev-1.13.0-intent-classification-core","2026-10-core-v1-jev-1.13.0-oos-calibration-core"],"deployment":"api","effective_model_id":"jev-1.13.0","execution_count":2,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3","machine":"arm64","release":"25.6.0","run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":0,"failure_rate":0.0,"family":"jev","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":1062302,"known_provider_cost_quality_pareto":true,"latency_mean_ms":258.10247596585555,"latency_p50_ms":248.95504149753833,"latency_p95_ms":315.9772399922076,"latest_completed_at_utc":"2026-10-06T12:51:50.754874+00:00","model_id":"jev-1.13.0","model_key":"jev-1.13.0","model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","observed_case_count":616,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":true,"observed_quality_parameters_pareto":false,"output_tokens_total":511838,"overall_quality_score":86.03896103896105,"parameters_b":null,"provider_cost_coverage_rate":1.0,"provider_cost_known":true,"provider_cost_observed_total_usd":0.044616684000000004,"provider_cost_per_1k_cases_usd":0.07242968181818182,"provider_cost_per_case_usd":7.242968181818182e-05,"provider_cost_priced_cases":616,"provider_cost_pricing":{"as_of":"2026-09-20","currency":"USD","price_key":"jev","processing":"standard","source":"TypeSafe AI public pricing: $42 per billion input tokens; output free.","source_url":"[redacted local or credential-bearing value]"},"provider_cost_status":"complete","provider_cost_total_cases":616,"provider_cost_total_usd":0.044616684000000004,"provider_key":"typesafe","quality_coverage_capabilities":2,"quality_coverage_complete":false,"quality_coverage_datasets":2,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":null,"resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":null,"peak_execution_signature":null,"peak_run_id":null,"process_cpu_percent_avg":null,"process_rss_bytes_avg":null,"process_rss_bytes_peak":null,"resource_cell_count":0,"sample_count":0,"sampling_error_count":0,"scope":"unavailable","source":"unavailable","system_available_memory_bytes_min":null},"result_policy":"standard","runtime_key":"typesafe-api","tags":["api","closed","typesafe","decision-model"]},{"artifact_format":"gguf","artifact_size_bytes":10947100928,"current_run_ids":["lfm25-consolidated-latest","lfm25-math-v2-core"],"deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_count":2,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:503d302009e21bb18588b87ac3897f9112994a479905bbca872384efa4360298","machine":"arm64","release":"25.6.0","run_id":"lfm25-math-v2-core","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":15,"failure_rate":0.02952755905511811,"family":"lfm2.5","generation_parameters":{"extra":{},"max_output_tokens":2048,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":223016,"known_provider_cost_quality_pareto":false,"latency_mean_ms":16482.3864442058,"latency_p50_ms":9803.956041461788,"latency_p95_ms":60905.82317115038,"latest_completed_at_utc":"2026-10-22T14:07:00.510877+00:00","model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","observed_case_count":508,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":true,"output_tokens_total":373453,"overall_quality_score":85.23629148629148,"parameters_b":24.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":508,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"UD-Q3_K_XL","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:503d302009e21bb18588b87ac3897f9112994a479905bbca872384efa4360298","machine":"arm64","release":"25.6.0","run_id":"lfm25-math-v2-core","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:503d302009e21bb18588b87ac3897f9112994a479905bbca872384efa4360298","peak_run_id":"lfm25-math-v2-core","process_cpu_percent_avg":10.29314878109045,"process_rss_bytes_avg":1417688467.7424607,"process_rss_bytes_peak":1463648256.0,"resource_cell_count":1,"sample_count":1316,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4090920960.0},"result_policy":"consolidated-latest-attempt","runtime_key":"korgis-local","tags":["local","open-weight","moe"]},{"artifact_format":null,"artifact_size_bytes":null,"current_run_ids":["16a842b0-5dcc-47b5-8808-850b2ad1ac10","8eef4bab-617a-4638-9e51-53e72034e821","b6b4123a-1d20-4c39-8644-ea00c1e5893b"],"deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_count":3,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","machine":"arm64","release":"25.6.0","run_id":"b6b4123a-1d20-4c39-8644-ea00c1e5893b","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"smoke","failure_count":14,"failure_rate":0.15217391304347827,"family":"limite","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":5563,"known_provider_cost_quality_pareto":false,"latency_mean_ms":19307.593355981542,"latency_p50_ms":14347.055187972728,"latency_p95_ms":51331.97907336871,"latest_completed_at_utc":"2026-10-20T10:03:44.604674+00:00","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","observed_case_count":92,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":true,"output_tokens_total":71678,"overall_quality_score":56.666666666666664,"parameters_b":1.0,"provider_cost_coverage_rate":0.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":0,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":92,"provider_cost_total_usd":null,"provider_key":"transformers","quality_coverage_capabilities":3,"quality_coverage_complete":false,"quality_coverage_datasets":3,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":null,"resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":null,"peak_execution_signature":null,"peak_run_id":null,"process_cpu_percent_avg":null,"process_rss_bytes_avg":null,"process_rss_bytes_peak":null,"resource_cell_count":0,"sample_count":0,"sampling_error_count":0,"scope":"unavailable","source":"unavailable","system_available_memory_bytes_min":null},"result_policy":"standard","runtime_key":"transformers-local","tags":["local","open-weight","limite","violetto","mathematics","reasoning","small","1b","in-process"]},{"artifact_format":"safetensors","artifact_size_bytes":416,"current_run_ids":["d445eb9f-1c5a-4781-83be-43b6eeee3a44"],"deployment":"local","effective_model_id":"limite-1b-violetto-mlx","execution_count":1,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:f4c681453ad106c8c878d601ee4a0ee6335674e81abaf3f7ba5297ce35841db0","machine":"arm64","release":"25.6.0","run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"smoke","failure_count":10,"failure_rate":1.0,"family":"limite","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":null,"known_provider_cost_quality_pareto":false,"latency_mean_ms":21810.495604295284,"latency_p50_ms":21736.863624944817,"latency_p95_ms":22921.96444603032,"latest_completed_at_utc":"2026-10-20T20:00:36.317482+00:00","model_id":"gabe-rbo/limite-1b-violetto-mlx","model_key":"limite-1b-violetto-mlx","model_signature":"sha256:model:fddf01c0845257f40bbb635a92054df8eb22dacd450b827a951962e906d6513b","observed_case_count":10,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":null,"overall_quality_score":0.0,"parameters_b":1.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":10,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":10,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":1,"quality_coverage_complete":false,"quality_coverage_datasets":1,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":null,"resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:f4c681453ad106c8c878d601ee4a0ee6335674e81abaf3f7ba5297ce35841db0","machine":"arm64","release":"25.6.0","run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:f4c681453ad106c8c878d601ee4a0ee6335674e81abaf3f7ba5297ce35841db0","peak_run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","process_cpu_percent_avg":54.47089950027608,"process_rss_bytes_avg":96917494.20591466,"process_rss_bytes_peak":102842368.0,"resource_cell_count":1,"sample_count":670,"sampling_error_count":0,"scope":"server_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":7568949248.0},"result_policy":"standard","runtime_key":"korgis-local","tags":["local","open-weight","limite","violetto","mathematics","reasoning","small","1b","mlx","apple-silicon","korgis"]},{"artifact_format":"gguf","artifact_size_bytes":null,"current_run_ids":["2026-10-core-v1-nemotron-nano-4b-intent-classification-core","2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core","2026-10-core-v1-nemotron-nano-4b-oos-calibration-core","2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","2026-10-core-v1-nemotron-nano-4b-structured-output-core","3b0292d4-8f8f-4843-900c-946c32476964"],"deployment":"local","effective_model_id":"nemotron-nano-4b","execution_count":6,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:0caf2fff221781464037ba78ff0ed5a8158d632ab0e66481fc8df733dd8785c6","machine":"arm64","release":"25.6.0","run_id":"3b0292d4-8f8f-4843-900c-946c32476964","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":2,"failure_rate":0.003937007874015748,"family":"nemotron-nano","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":228654,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1742.2881335475013,"latency_p50_ms":1999.5908125056303,"latency_p95_ms":2575.8644812565763,"latest_completed_at_utc":"2026-10-15T15:45:29.838316+00:00","model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","observed_case_count":508,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":true,"output_tokens_total":8779,"overall_quality_score":71.57467532467533,"parameters_b":4,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":508,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","machine":"arm64","release":"25.6.0","run_id":"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","peak_run_id":"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","process_cpu_percent_avg":8.364117435376155,"process_rss_bytes_avg":5701263391.844644,"process_rss_bytes_peak":11089903616.0,"resource_cell_count":6,"sample_count":3679,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5000953856.0},"result_policy":"standard","runtime_key":"korgis-local","tags":["local","open-weight","nemotron","small"]},{"artifact_format":"gguf","artifact_size_bytes":null,"current_run_ids":["2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core","2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core","2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core","2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core","b10ad834-2bdd-409b-9073-49f0233792ea"],"deployment":"local","effective_model_id":"nemotron-nano-4b-q8","execution_count":6,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:89a74e35df26090fe2c6adb16b01e32373fbcdad6966a1b23cc9b88ba35a5170","machine":"arm64","release":"25.6.0","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":2,"failure_rate":0.003937007874015748,"family":"nemotron-nano","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":228654,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2567.452957344186,"latency_p50_ms":2537.835583498236,"latency_p95_ms":4611.300510846194,"latest_completed_at_utc":"2026-10-16T08:53:37.155498+00:00","model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","observed_case_count":508,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":8897,"overall_quality_score":71.18867243867244,"parameters_b":4,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":508,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","machine":"arm64","release":"25.6.0","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","peak_run_id":"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","process_cpu_percent_avg":6.5230274878355,"process_rss_bytes_avg":6466281556.123229,"process_rss_bytes_peak":11438456832.0,"resource_cell_count":6,"sample_count":4868,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5441241088.0},"result_policy":"standard","runtime_key":"korgis-local","tags":["local","open-weight","nemotron","medium"]},{"artifact_format":"gguf","artifact_size_bytes":null,"current_run_ids":["2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core","2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core","2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core","2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core","b10ad834-2bdd-409b-9073-49f0233792ea"],"deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","execution_count":6,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:c58f103ecf1090aab4510c6708cd9446fefc4137c61ae9ca33da8cc1c5d3faf8","machine":"arm64","release":"25.6.0","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":3,"failure_rate":0.005905511811023622,"family":"qwen3.5","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":204788,"known_provider_cost_quality_pareto":false,"latency_mean_ms":344.54570039366,"latency_p50_ms":347.7319170051487,"latency_p95_ms":474.87177490329475,"latest_completed_at_utc":"2026-10-16T08:53:37.155498+00:00","model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","observed_case_count":508,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":true,"output_tokens_total":8771,"overall_quality_score":37.43145743145743,"parameters_b":0.8,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":508,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","machine":"arm64","release":"25.6.0","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","peak_run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","process_cpu_percent_avg":16.61227127427062,"process_rss_bytes_avg":3002504253.7158256,"process_rss_bytes_peak":9496838144.0,"resource_cell_count":6,"sample_count":1407,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5715886080.0},"result_policy":"standard","runtime_key":"korgis-local","tags":["local","open-weight","qwen","small"]},{"artifact_format":"gguf","artifact_size_bytes":3775709216,"current_run_ids":["b10ad834-2bdd-409b-9073-49f0233792ea","qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core","qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core","qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core","qwen-variants-core-qwen3.5-2b-bf16-structured-output-core"],"deployment":"local","effective_model_id":"qwen3.5-2b-bf16","execution_count":6,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","machine":"arm64","release":"25.6.0","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":15,"failure_rate":0.02952755905511811,"family":"qwen3.5","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":204788,"known_provider_cost_quality_pareto":false,"latency_mean_ms":902.998680881693,"latency_p50_ms":822.7691460051574,"latency_p95_ms":1520.380753971403,"latest_completed_at_utc":"2026-10-16T08:53:37.155498+00:00","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","observed_case_count":508,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":8763,"overall_quality_score":50.88203463203464,"parameters_b":2.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":508,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"BF16","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","machine":"arm64","release":"25.6.0","run_id":"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","peak_run_id":"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","process_cpu_percent_avg":6.167834376927058,"process_rss_bytes_avg":5043064968.8305025,"process_rss_bytes_peak":9291366400.0,"resource_cell_count":6,"sample_count":2246,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4923883520.0},"result_policy":"standard","runtime_key":"korgis-local","tags":["local","open-weight","qwen","small","bf16"]},{"artifact_format":"gguf","artifact_size_bytes":null,"current_run_ids":["2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core","2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core","2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core","2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core","9fb9f932-1f29-4646-b337-f33476d841ba"],"deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_count":6,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:8190c8f3bb1f7b51ef597577c913ceae05f8cb7998a7d5221e848f67071166c9","machine":"arm64","release":"25.6.0","run_id":"9fb9f932-1f29-4646-b337-f33476d841ba","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":4,"failure_rate":0.007874015748031496,"family":"qwen3.5","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":204788,"known_provider_cost_quality_pareto":false,"latency_mean_ms":646.8228120271721,"latency_p50_ms":700.4097294993699,"latency_p95_ms":879.9385391488613,"latest_completed_at_utc":"2026-10-15T15:44:53.050810+00:00","model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","observed_case_count":508,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":8820,"overall_quality_score":52.826479076479075,"parameters_b":2,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":508,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","machine":"arm64","release":"25.6.0","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","peak_run_id":"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","process_cpu_percent_avg":11.662933984950136,"process_rss_bytes_avg":2370946166.6539683,"process_rss_bytes_peak":6107152384.0,"resource_cell_count":6,"sample_count":1870,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5225758720.0},"result_policy":"standard","runtime_key":"korgis-local","tags":["local","open-weight","qwen","small"]},{"artifact_format":"gguf","artifact_size_bytes":null,"current_run_ids":["3a46714c-d579-4688-a8b0-558183b4e02a","b10ad834-2bdd-409b-9073-49f0233792ea"],"deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_count":2,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:d6ef7d6b6d33f70d69c5d1df4568941686a610cda0bde994801472a902500dd4","machine":"arm64","release":"25.6.0","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":3,"failure_rate":0.005905511811023622,"family":"qwen3.5","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":204788,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1429.9514306818367,"latency_p50_ms":1585.6053334937315,"latency_p95_ms":1972.186656796839,"latest_completed_at_utc":"2026-10-16T08:53:37.155498+00:00","model_id":"unsloth/qwen3.5-4b","model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","observed_case_count":508,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":8520,"overall_quality_score":68.63095238095238,"parameters_b":4,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":508,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","machine":"arm64","release":"25.6.0","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","peak_run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","process_cpu_percent_avg":9.008174663870694,"process_rss_bytes_avg":5701665635.721027,"process_rss_bytes_peak":10744905728.0,"resource_cell_count":6,"sample_count":3195,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5199380480.0},"result_policy":"standard","runtime_key":"korgis-local","tags":["local","open-weight","qwen","4b"]},{"artifact_format":"gguf","artifact_size_bytes":4482403488,"current_run_ids":["b10ad834-2bdd-409b-9073-49f0233792ea","qwen-variants-core-qwen3.5-4b-q8-intent-classification-core","qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core","qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core","qwen-variants-core-qwen3.5-4b-q8-structured-output-core"],"deployment":"local","effective_model_id":"qwen3.5-4b-q8","execution_count":6,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","machine":"arm64","release":"25.6.0","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":2,"failure_rate":0.003937007874015748,"family":"qwen3.5","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":204788,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1605.0692064233635,"latency_p50_ms":1699.8242709960323,"latency_p95_ms":2421.302295583882,"latest_completed_at_utc":"2026-10-16T08:53:37.155498+00:00","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","observed_case_count":508,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":8488,"overall_quality_score":70.08116883116882,"parameters_b":4.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":508,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"Q8_0","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","machine":"arm64","release":"25.6.0","run_id":"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","peak_run_id":"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","process_cpu_percent_avg":6.715570632877846,"process_rss_bytes_avg":6928149991.082301,"process_rss_bytes_peak":11637391360.0,"resource_cell_count":6,"sample_count":3285,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4841750528.0},"result_policy":"standard","runtime_key":"korgis-local","tags":["local","open-weight","qwen","4b","q8"]},{"artifact_format":"gguf","artifact_size_bytes":null,"current_run_ids":["2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core","2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core","2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core","2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core","b10ad834-2bdd-409b-9073-49f0233792ea"],"deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_count":6,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:9d728f687c9bb0f47a2c5a8c0e61b6da5a7e2833ad1afcffb18c7b4eb2767886","machine":"arm64","release":"25.6.0","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":2,"failure_rate":0.003937007874015748,"family":"qwen3.5","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":204788,"known_provider_cost_quality_pareto":false,"latency_mean_ms":3119.1373101190284,"latency_p50_ms":2776.106687495485,"latency_p95_ms":7684.784721046889,"latest_completed_at_utc":"2026-10-16T08:53:37.155498+00:00","model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","observed_case_count":508,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":true,"output_tokens_total":8355,"overall_quality_score":75.73051948051948,"parameters_b":9,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":508,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","machine":"arm64","release":"25.6.0","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","peak_run_id":"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","process_cpu_percent_avg":5.385318471481629,"process_rss_bytes_avg":7625426284.848329,"process_rss_bytes_peak":11538563072.0,"resource_cell_count":6,"sample_count":5781,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5142609920.0},"result_policy":"standard","runtime_key":"korgis-local","tags":["local","open-weight","qwen"]},{"artifact_format":"gguf","artifact_size_bytes":4121781472,"current_run_ids":["qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core","qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core","qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core","qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core"],"deployment":"local","effective_model_id":"qwen3.5-9b-ud-q2kxl","execution_count":5,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","machine":"arm64","release":"25.6.0","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":4,"failure_rate":0.008547008547008548,"family":"qwen3.5","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":200390,"known_provider_cost_quality_pareto":false,"latency_mean_ms":2500.572187915712,"latency_p50_ms":2746.6233744926285,"latency_p95_ms":3354.4908000039864,"latest_completed_at_utc":"2026-10-13T12:04:10.873473+00:00","model_id":"qwen/qwen3.5-9b-ud-q2kxl","model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","observed_case_count":468,"observed_quality_artifact_size_pareto":true,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":8167,"overall_quality_score":75.65367965367966,"parameters_b":9.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":468,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":468,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":5,"quality_coverage_complete":false,"quality_coverage_datasets":5,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"UD-Q2_K_XL","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","machine":"arm64","release":"25.6.0","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","peak_run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","process_cpu_percent_avg":4.278282862702705,"process_rss_bytes_avg":7033561782.88888,"process_rss_bytes_peak":11224416256.0,"resource_cell_count":5,"sample_count":4311,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5265096704.0},"result_policy":"standard","runtime_key":"korgis-local","tags":["local","open-weight","qwen","9b","q2"]},{"artifact_format":"gguf","artifact_size_bytes":6192222208,"current_run_ids":["2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core"],"deployment":"local","effective_model_id":"qwen3.8-27b-iq1s","execution_count":5,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","machine":"arm64","release":"25.6.0","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":36,"failure_rate":0.07692307692307693,"family":"qwen3.8","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":189885,"known_provider_cost_quality_pareto":false,"latency_mean_ms":8807.17722295277,"latency_p50_ms":8610.94677100482,"latency_p95_ms":12756.54564594878,"latest_completed_at_utc":"2026-10-06T14:53:01.287222+00:00","model_id":"Qwen/Qwen3.8-27B","model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","observed_case_count":468,"observed_quality_artifact_size_pareto":false,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":7958,"overall_quality_score":62.71861471861472,"parameters_b":27.0,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":468,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":468,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":5,"quality_coverage_complete":false,"quality_coverage_datasets":5,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"UD-IQ1_S","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","machine":"arm64","release":"25.6.0","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","peak_run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","process_cpu_percent_avg":5.0778736832568985,"process_rss_bytes_avg":8497387609.154477,"process_rss_bytes_peak":13159825408.0,"resource_cell_count":5,"sample_count":13875,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4307910656.0},"result_policy":"standard","runtime_key":"korgis-local","tags":["local","open-weight","qwen","reasoning","iq1s","27b"]},{"artifact_format":"gguf","artifact_size_bytes":2600224352,"current_run_ids":["2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core","2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core","2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core","2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core","e68754ab-25e6-4449-b06a-66dbc311d9e5"],"deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_count":6,"execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:d26122294632fb1ef7e62445eece13d61030562e5a3d7a64b60606c3c12ff3be","machine":"arm64","release":"25.6.0","run_id":"e68754ab-25e6-4449-b06a-66dbc311d9e5","system":"darwin","total_memory_bytes":38654705664},"execution_profile":"core","failure_count":1,"failure_rate":0.001968503937007874,"family":"spark-x2.5","generation_parameters":{"extra":{},"max_output_tokens":256,"seed":42,"stop":[],"temperature":0.0},"input_tokens_total":234269,"known_provider_cost_quality_pareto":false,"latency_mean_ms":1776.7870538910634,"latency_p50_ms":2137.5340625018,"latency_p95_ms":2707.341406807245,"latest_completed_at_utc":"2026-10-15T15:45:57.052259+00:00","model_id":"xhtoken/spark-x2.5-4b","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","observed_case_count":508,"observed_quality_artifact_size_pareto":true,"observed_quality_latency_pareto":false,"observed_quality_parameters_pareto":false,"output_tokens_total":9342,"overall_quality_score":67.95093795093796,"parameters_b":4,"provider_cost_coverage_rate":1.0,"provider_cost_known":false,"provider_cost_observed_total_usd":null,"provider_cost_per_1k_cases_usd":null,"provider_cost_per_case_usd":null,"provider_cost_priced_cases":508,"provider_cost_pricing":null,"provider_cost_status":"local_not_applicable","provider_cost_total_cases":508,"provider_cost_total_usd":null,"provider_key":"korgis","quality_coverage_capabilities":6,"quality_coverage_complete":true,"quality_coverage_datasets":6,"quality_policy_id":"core-quality-v1","quality_total_capabilities":6,"quality_total_datasets":6,"quantization":"Q4_K_M","resource_summary":{"accelerator_memory_bytes_peak":null,"peak_execution_environment":{"cpu_model":"Apple M3 Pro","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","machine":"arm64","release":"25.6.0","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","system":"darwin","total_memory_bytes":38654705664},"peak_execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","peak_run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","process_cpu_percent_avg":7.0819593917060635,"process_rss_bytes_avg":3775214581.6258616,"process_rss_bytes_peak":7862026240.0,"resource_cell_count":6,"sample_count":3654,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5815320576.0},"result_policy":"standard","runtime_key":"korgis-local","tags":["local","open-weight","spark","reasoning","coding","agentic","q4","4b"]}],"quality_policy":{"aggregation":"mean_normalized_primary_metric","coverage_requirement":"complete","label":"Equal-weight capability quality","normalization":"higher-is-better unit metrics mapped to 0-100","quality_policy_id":"core-quality-v1"},"schema_version":"2"},"frontier":{"compression_groups":["bonsai:27b","lfm2.5:24b","limite:1b","nemotron-nano:4b","qwen3.5:0.8b","qwen3.5:2b","qwen3.5:4b","qwen3.5:9b","qwen3.8:27b","spark-x2.5:4b"],"families":["bonsai","gemini-3.5","gemini-3.8","gpt-5.6","gpt-6","jev","lfm2.5","limite","nemotron-nano","qwen3.5","qwen3.8","spark-x2.5"],"points":[{"artifact_size_bytes":null,"compression_group":"bonsai:27b","deployment":"local","family":"bonsai","latency_p50_ms":8354.460334005125,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","parameters_b":27.0,"pareto_artifact_size":false,"pareto_latency":false,"pareto_parameters":false,"pareto_peak_rss":false,"peak_rss_bytes":11685068800.0,"quality":70.75757575757575,"quantization":"Q1_0"},{"artifact_size_bytes":null,"compression_group":"qwen3.5:2b","deployment":"local","family":"qwen3.5","latency_p50_ms":1149.7726044981391,"model_key":"decisio-qwen3.5-2b-q4km","model_signature":"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95","parameters_b":2.0,"pareto_artifact_size":false,"pareto_latency":false,"pareto_parameters":true,"pareto_peak_rss":false,"peak_rss_bytes":null,"quality":60.3896103896104,"quantization":"Q4_K_M"},{"artifact_size_bytes":null,"compression_group":"qwen3.5:4b","deployment":"local","family":"qwen3.5","latency_p50_ms":2828.9587710023625,"model_key":"decisio-qwen3.5-4b-q4km","model_signature":"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a","parameters_b":4.0,"pareto_artifact_size":false,"pareto_latency":false,"pareto_parameters":false,"pareto_peak_rss":false,"peak_rss_bytes":null,"quality":70.12987012987013,"quantization":"Q4_K_M"},{"artifact_size_bytes":null,"compression_group":"qwen3.5:9b","deployment":"local","family":"qwen3.5","latency_p50_ms":4986.015040994971,"model_key":"decisio-qwen3.5-9b-q4km","model_signature":"sha256:model:ae0099eeb8925bc828c92072ba2f486986899854d2d3de76e5689cd383c783e3","parameters_b":9.0,"pareto_artifact_size":false,"pareto_latency":false,"pareto_parameters":false,"pareto_peak_rss":false,"peak_rss_bytes":null,"quality":66.23376623376623,"quantization":"Q4_K_M"},{"artifact_size_bytes":null,"compression_group":null,"deployment":"api","family":"gemini-3.5","latency_p50_ms":133.2621249894146,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","parameters_b":null,"pareto_artifact_size":false,"pareto_latency":true,"pareto_parameters":false,"pareto_peak_rss":false,"peak_rss_bytes":null,"quality":11.25,"quantization":null},{"artifact_size_bytes":null,"compression_group":null,"deployment":"api","family":"gemini-3.8","latency_p50_ms":129.50879096752033,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","parameters_b":null,"pareto_artifact_size":false,"pareto_latency":true,"pareto_parameters":false,"pareto_peak_rss":false,"peak_rss_bytes":null,"quality":10.0,"quantization":null},{"artifact_size_bytes":null,"compression_group":null,"deployment":"api","family":"gpt-5.6","latency_p50_ms":1445.532062505663,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","parameters_b":null,"pareto_artifact_size":false,"pareto_latency":false,"pareto_parameters":false,"pareto_peak_rss":false,"peak_rss_bytes":null,"quality":78.88347763347763,"quantization":null},{"artifact_size_bytes":null,"compression_group":null,"deployment":"api","family":"gpt-6","latency_p50_ms":1896.0361664940137,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","parameters_b":null,"pareto_artifact_size":false,"pareto_latency":false,"pareto_parameters":false,"pareto_peak_rss":false,"peak_rss_bytes":null,"quality":78.3423520923521,"quantization":null},{"artifact_size_bytes":null,"compression_group":null,"deployment":"api","family":"jev","latency_p50_ms":248.95504149753833,"model_key":"jev-1.13.0","model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","parameters_b":null,"pareto_artifact_size":false,"pareto_latency":true,"pareto_parameters":false,"pareto_peak_rss":false,"peak_rss_bytes":null,"quality":86.03896103896105,"quantization":null},{"artifact_size_bytes":10947100928,"compression_group":"lfm2.5:24b","deployment":"local","family":"lfm2.5","latency_p50_ms":9803.956041461788,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"pareto_artifact_size":true,"pareto_latency":false,"pareto_parameters":true,"pareto_peak_rss":true,"peak_rss_bytes":1463648256.0,"quality":85.23629148629148,"quantization":"UD-Q3_K_XL"},{"artifact_size_bytes":null,"compression_group":"limite:1b","deployment":"local","family":"limite","latency_p50_ms":14347.055187972728,"model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","parameters_b":1.0,"pareto_artifact_size":false,"pareto_latency":false,"pareto_parameters":true,"pareto_peak_rss":false,"peak_rss_bytes":null,"quality":56.666666666666664,"quantization":null},{"artifact_size_bytes":416,"compression_group":"limite:1b","deployment":"local","family":"limite","latency_p50_ms":21736.863624944817,"model_key":"limite-1b-violetto-mlx","model_signature":"sha256:model:fddf01c0845257f40bbb635a92054df8eb22dacd450b827a951962e906d6513b","parameters_b":1.0,"pareto_artifact_size":true,"pareto_latency":false,"pareto_parameters":false,"pareto_peak_rss":true,"peak_rss_bytes":102842368.0,"quality":0.0,"quantization":null},{"artifact_size_bytes":null,"compression_group":"nemotron-nano:4b","deployment":"local","family":"nemotron-nano","latency_p50_ms":1999.5908125056303,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","parameters_b":4.0,"pareto_artifact_size":false,"pareto_latency":false,"pareto_parameters":true,"pareto_peak_rss":false,"peak_rss_bytes":11089903616.0,"quality":71.57467532467533,"quantization":"Q4_K_M"},{"artifact_size_bytes":null,"compression_group":"nemotron-nano:4b","deployment":"local","family":"nemotron-nano","latency_p50_ms":2537.835583498236,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","parameters_b":4.0,"pareto_artifact_size":false,"pareto_latency":false,"pareto_parameters":false,"pareto_peak_rss":false,"peak_rss_bytes":11438456832.0,"quality":71.18867243867244,"quantization":"Q8_0"},{"artifact_size_bytes":null,"compression_group":"qwen3.5:0.8b","deployment":"local","family":"qwen3.5","latency_p50_ms":347.7319170051487,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","parameters_b":0.8,"pareto_artifact_size":false,"pareto_latency":false,"pareto_parameters":true,"pareto_peak_rss":false,"peak_rss_bytes":9496838144.0,"quality":37.43145743145743,"quantization":"Q4_K_M"},{"artifact_size_bytes":3775709216,"compression_group":"qwen3.5:2b","deployment":"local","family":"qwen3.5","latency_p50_ms":822.7691460051574,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","parameters_b":2.0,"pareto_artifact_size":false,"pareto_latency":false,"pareto_parameters":false,"pareto_peak_rss":false,"peak_rss_bytes":9291366400.0,"quality":50.88203463203464,"quantization":"BF16"},{"artifact_size_bytes":null,"compression_group":"qwen3.5:2b","deployment":"local","family":"qwen3.5","latency_p50_ms":700.4097294993699,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","parameters_b":2.0,"pareto_artifact_size":false,"pareto_latency":false,"pareto_parameters":false,"pareto_peak_rss":false,"peak_rss_bytes":6107152384.0,"quality":52.826479076479075,"quantization":"Q4_K_M"},{"artifact_size_bytes":null,"compression_group":"qwen3.5:4b","deployment":"local","family":"qwen3.5","latency_p50_ms":1585.6053334937315,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","parameters_b":4.0,"pareto_artifact_size":false,"pareto_latency":false,"pareto_parameters":false,"pareto_peak_rss":false,"peak_rss_bytes":10744905728.0,"quality":68.63095238095238,"quantization":"Q4_K_M"},{"artifact_size_bytes":4482403488,"compression_group":"qwen3.5:4b","deployment":"local","family":"qwen3.5","latency_p50_ms":1699.8242709960323,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","parameters_b":4.0,"pareto_artifact_size":false,"pareto_latency":false,"pareto_parameters":false,"pareto_peak_rss":false,"peak_rss_bytes":11637391360.0,"quality":70.08116883116882,"quantization":"Q8_0"},{"artifact_size_bytes":null,"compression_group":"qwen3.5:9b","deployment":"local","family":"qwen3.5","latency_p50_ms":2776.106687495485,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","parameters_b":9.0,"pareto_artifact_size":false,"pareto_latency":false,"pareto_parameters":true,"pareto_peak_rss":false,"peak_rss_bytes":11538563072.0,"quality":75.73051948051948,"quantization":"Q4_K_M"},{"artifact_size_bytes":4121781472,"compression_group":"qwen3.5:9b","deployment":"local","family":"qwen3.5","latency_p50_ms":2746.6233744926285,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","parameters_b":9.0,"pareto_artifact_size":true,"pareto_latency":false,"pareto_parameters":false,"pareto_peak_rss":false,"peak_rss_bytes":11224416256.0,"quality":75.65367965367966,"quantization":"UD-Q2_K_XL"},{"artifact_size_bytes":6192222208,"compression_group":"qwen3.8:27b","deployment":"local","family":"qwen3.8","latency_p50_ms":8610.94677100482,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","parameters_b":27.0,"pareto_artifact_size":false,"pareto_latency":false,"pareto_parameters":false,"pareto_peak_rss":false,"peak_rss_bytes":13159825408.0,"quality":62.71861471861472,"quantization":"UD-IQ1_S"},{"artifact_size_bytes":2600224352,"compression_group":"spark-x2.5:4b","deployment":"local","family":"spark-x2.5","latency_p50_ms":2137.5340625018,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","parameters_b":4.0,"pareto_artifact_size":true,"pareto_latency":false,"pareto_parameters":false,"pareto_peak_rss":false,"peak_rss_bytes":7862026240.0,"quality":67.95093795093796,"quantization":"Q4_K_M"}],"schema_version":"1"},"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"bonsai-27b-q10","family":"bonsai","model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","parameters_b":27,"provider_key":"korgis","quantization":"Q1_0","runtime_key":"korgis-local","tags":["local","open-weight","bonsai","27b"]},{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-2b-q4km","family":"qwen3.5","model_id":"unsloth/qwen3.5-2b","model_key":"decisio-qwen3.5-2b-q4km","model_signature":"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95","parameters_b":2,"provider_key":"decisio","quantization":"Q4_K_M","runtime_key":"decisio-local","tags":["local","decisio","open-weight","qwen","decision-model"]},{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-4b-q4km","family":"qwen3.5","model_id":"unsloth/qwen3.5-4b","model_key":"decisio-qwen3.5-4b-q4km","model_signature":"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a","parameters_b":4,"provider_key":"decisio","quantization":"Q4_K_M","runtime_key":"decisio-local","tags":["local","decisio","open-weight","qwen","decision-model"]},{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-9b-q4km","family":"qwen3.5","model_id":"qwen/qwen3.5-9b","model_key":"decisio-qwen3.5-9b-q4km","model_signature":"sha256:model:ae0099eeb8925bc828c92072ba2f486986899854d2d3de76e5689cd383c783e3","parameters_b":9,"provider_key":"decisio","quantization":"Q4_K_M","runtime_key":"decisio-local","tags":["local","decisio","open-weight","qwen","decision-model"]},{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"gemini-3.5-flash","family":"gemini-3.5","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","parameters_b":null,"provider_key":"gemini","quantization":null,"runtime_key":"gemini-api","tags":["api","closed","google","gemini","flash"]},{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"gemini-3.8-flash","family":"gemini-3.8","model_id":"gemini-3.8-flash","model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","parameters_b":null,"provider_key":"gemini","quantization":null,"runtime_key":"gemini-api","tags":["api","closed","google","gemini","flash"]},{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"gpt-5.6-luna","family":"gpt-5.6","model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","parameters_b":null,"provider_key":"openai","quantization":null,"runtime_key":"openai-api","tags":["api","closed","openai"]},{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"gpt-6-luna","family":"gpt-6","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","parameters_b":null,"provider_key":"openai","quantization":null,"runtime_key":"openai-api","tags":["api","closed","openai"]},{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"jev-1.13.0","family":"jev","model_id":"jev-1.13.0","model_key":"jev-1.13.0","model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","parameters_b":null,"provider_key":"typesafe","quantization":null,"runtime_key":"typesafe-api","tags":["api","closed","typesafe","decision-model"]},{"artifact_format":"gguf","artifact_size_bytes":10947100928,"deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","family":"lfm2.5","model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"provider_key":"korgis","quantization":"UD-Q3_K_XL","runtime_key":"korgis-local","tags":["local","open-weight","moe"]},{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"nemotron-nano-4b","family":"nemotron-nano","model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","parameters_b":4,"provider_key":"korgis","quantization":"Q4_K_M","runtime_key":"korgis-local","tags":["local","open-weight","nemotron","small"]},{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"nemotron-nano-4b-q8","family":"nemotron-nano","model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","parameters_b":4,"provider_key":"korgis","quantization":"Q8_0","runtime_key":"korgis-local","tags":["local","open-weight","nemotron","medium"]},{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","family":"qwen3.5","model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","parameters_b":0.8,"provider_key":"korgis","quantization":"Q4_K_M","runtime_key":"korgis-local","tags":["local","open-weight","qwen","small"]},{"artifact_format":"gguf","artifact_size_bytes":3775709216,"deployment":"local","effective_model_id":"qwen3.5-2b-bf16","family":"qwen3.5","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","parameters_b":2.0,"provider_key":"korgis","quantization":"BF16","runtime_key":"korgis-local","tags":["local","open-weight","qwen","small","bf16"]},{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-2b-q4km","family":"qwen3.5","model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","parameters_b":2,"provider_key":"korgis","quantization":"Q4_K_M","runtime_key":"korgis-local","tags":["local","open-weight","qwen","small"]},{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-4b-q4km","family":"qwen3.5","model_id":"unsloth/qwen3.5-4b","model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","parameters_b":4,"provider_key":"korgis","quantization":"Q4_K_M","runtime_key":"korgis-local","tags":["local","open-weight","qwen","4b"]},{"artifact_format":"gguf","artifact_size_bytes":4482403488,"deployment":"local","effective_model_id":"qwen3.5-4b-q8","family":"qwen3.5","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","parameters_b":4.0,"provider_key":"korgis","quantization":"Q8_0","runtime_key":"korgis-local","tags":["local","open-weight","qwen","4b","q8"]},{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-9b-q4km","family":"qwen3.5","model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","parameters_b":9,"provider_key":"korgis","quantization":"Q4_K_M","runtime_key":"korgis-local","tags":["local","open-weight","qwen"]},{"artifact_format":"gguf","artifact_size_bytes":4121781472,"deployment":"local","effective_model_id":"qwen3.5-9b-ud-q2kxl","family":"qwen3.5","model_id":"qwen/qwen3.5-9b-ud-q2kxl","model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","parameters_b":9.0,"provider_key":"korgis","quantization":"UD-Q2_K_XL","runtime_key":"korgis-local","tags":["local","open-weight","qwen","9b","q2"]},{"artifact_format":"gguf","artifact_size_bytes":6192222208,"deployment":"local","effective_model_id":"qwen3.8-27b-iq1s","family":"qwen3.8","model_id":"Qwen/Qwen3.8-27B","model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","parameters_b":27.0,"provider_key":"korgis","quantization":"UD-IQ1_S","runtime_key":"korgis-local","tags":["local","open-weight","qwen","reasoning","iq1s","27b"]},{"artifact_format":"gguf","artifact_size_bytes":2600224352,"deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","family":"spark-x2.5","model_id":"xhtoken/spark-x2.5-4b","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","parameters_b":4,"provider_key":"korgis","quantization":"Q4_K_M","runtime_key":"korgis-local","tags":["local","open-weight","spark","reasoning","coding","agentic","q4","4b"]},{"artifact_format":null,"artifact_size_bytes":null,"deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","family":"limite","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","parameters_b":1.0,"provider_key":"transformers","quantization":null,"runtime_key":"transformers-local","tags":["local","open-weight","limite","violetto","mathematics","reasoning","small","1b","in-process"]},{"artifact_format":null,"artifact_size_bytes":null,"deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","family":"limite","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65","parameters_b":1.0,"provider_key":"transformers","quantization":null,"runtime_key":"transformers-local","tags":["local","open-weight","limite","violetto","mathematics","reasoning","small","1b","in-process"]},{"artifact_format":"safetensors","artifact_size_bytes":416,"deployment":"local","effective_model_id":"limite-1b-violetto-mlx","family":"limite","model_id":"gabe-rbo/limite-1b-violetto-mlx","model_key":"limite-1b-violetto-mlx","model_signature":"sha256:model:fddf01c0845257f40bbb635a92054df8eb22dacd450b827a951962e906d6513b","parameters_b":1.0,"provider_key":"korgis","quantization":null,"runtime_key":"korgis-local","tags":["local","open-weight","limite","violetto","mathematics","reasoning","small","1b","mlx","apple-silicon","korgis"]},{"artifact_format":"gguf","artifact_size_bytes":2600224352,"deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","family":"spark-x2.5","model_id":"XHToken/Spark-X2.5-4B","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:8dcf43803761fd5570f637e166b655565b6c638e51a6a96848f58f6e6699a734","parameters_b":4.0,"provider_key":"korgis","quantization":"Q4_K_M","runtime_key":"korgis-local","tags":["local","open-weight","spark","reasoning","coding","agentic","q4","4b"]},{"artifact_format":null,"artifact_size_bytes":null,"deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","family":"limite","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:3fe1bfff5bb2c952f96bb799970eb76d9db17f9f880bc3326f0bdfbdd51577cc","parameters_b":1.0,"provider_key":"vllm","quantization":null,"runtime_key":"vllm-local","tags":["local","open-weight","limite","violetto","mathematics","reasoning","small","1b","in-process"]}],"result_view":"latest","runs":[{"completed_at_utc":"2026-10-22T14:07:00.510877+00:00","created_at_utc":"2026-10-22T13:59:54.129141+00:00","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","integrity_state":"VALID","manifest_checksum":"f088d8eeac7c8f9f978a001a29290a52b1e28f6143178cfb49cea1b6eeac8da3","profile":"core","run_group":"lfm25-math-v2","run_id":"lfm25-math-v2-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"3"},{"completed_at_utc":"2026-10-22T13:12:52.242438+00:00","created_at_utc":"2026-10-22T13:12:52.242438+00:00","git_commit":null,"integrity_state":"VALID","manifest_checksum":"2c2c5c7f7d18b2a1f0abf8ca525473967017f29964aac087ec0b38c00b6d1719","profile":"core","run_group":"lfm25-consolidated","run_id":"lfm25-consolidated-latest","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"3"},{"completed_at_utc":"2026-10-22T12:49:48.793583+00:00","created_at_utc":"2026-10-22T12:01:07.926009+00:00","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","integrity_state":"VALID","manifest_checksum":"146cea2f3b0118bec06dfff67ba54a9492d6c33e6a530a444d97be1acd6f00e6","profile":"core","run_group":"lfm25-recovery-4096","run_id":"lfm25-recovery-4096","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"3"},{"completed_at_utc":"2026-10-22T10:24:06.430109+00:00","created_at_utc":"2026-10-22T09:44:43.399412+00:00","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","integrity_state":"VALID","manifest_checksum":"adc478a4cc731d6468685075bb4fdc113e523a9d765a3209989619edeee45a61","profile":"core","run_group":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-oos-calibration-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"3"},{"completed_at_utc":"2026-10-22T09:44:42.382206+00:00","created_at_utc":"2026-10-22T08:51:39.298035+00:00","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","integrity_state":"VALID","manifest_checksum":"fd4427541a214849a942e2351a971ffa752ef9d68698969ea1d535e219ca67fb","profile":"core","run_group":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-intent-classification-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"3"},{"completed_at_utc":"2026-10-22T08:51:38.368018+00:00","created_at_utc":"2026-10-22T08:45:23.526981+00:00","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","integrity_state":"VALID","manifest_checksum":"6f0940e0a7d57020261afd6184bbdc2f72674291a065b58ecf6255ff891c5fe4","profile":"core","run_group":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-mathematical-reasoning-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"3"},{"completed_at_utc":"2026-10-22T08:45:22.525820+00:00","created_at_utc":"2026-10-22T08:16:15.433024+00:00","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","integrity_state":"VALID","manifest_checksum":"48bff421e9dafe622477d0662b2c0716ae488c39a0aad0140a4d4c6043f97834","profile":"core","run_group":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-qa-abstention-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"3"},{"completed_at_utc":"2026-10-22T08:16:14.436917+00:00","created_at_utc":"2026-10-22T07:56:31.502889+00:00","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","integrity_state":"VALID","manifest_checksum":"0cbce4ede0dba62fed1c64baa483a0c39a98e82608cb00881617dd1332fff6f5","profile":"core","run_group":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"3"},{"completed_at_utc":"2026-10-20T20:00:36.317482+00:00","created_at_utc":"2026-10-20T19:56:56.412647+00:00","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","integrity_state":"VALID","manifest_checksum":"9b1ae4f12731d5f65acf6095bef57ac4d0bc7712599fd1e9bd99de7562b83d8d","profile":"smoke","run_group":"limite-mlx-korgis-smoke-test","run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","seed":42,"source_dir":null,"status":"PARTIAL","suite_id":"capability-core","suite_version":"3"},{"completed_at_utc":"2026-10-20T12:33:29.825206+00:00","created_at_utc":"2026-10-20T12:33:29.776734+00:00","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","integrity_state":"VALID","manifest_checksum":"b48449dfb5de77769f39c1c03140956377f53f95d7883cdfb75dcfcf613fc6c4","profile":"smoke","run_group":"limite-violetto-q80-smoke","run_id":"ea0fd04b-aae3-4020-ac46-99d066560ecb","seed":42,"source_dir":null,"status":"PARTIAL","suite_id":"capability-core","suite_version":"3"},{"completed_at_utc":"2026-10-20T10:03:44.604674+00:00","created_at_utc":"2026-10-20T09:54:50.015691+00:00","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","integrity_state":"VALID","manifest_checksum":"ebc4ad6bec1c4798708fa4ce0f8383110ea6bbd3f84544c11dd2402600a4b295","profile":"smoke","run_group":"limite-violetto-qa-smoke","run_id":"b6b4123a-1d20-4c39-8644-ea00c1e5893b","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"3"},{"completed_at_utc":"2026-10-17T21:24:56.752644+00:00","created_at_utc":"2026-10-17T21:13:50.638554+00:00","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","integrity_state":"VALID","manifest_checksum":"6dfbf26de4ed844edc0871e5e69c778d861ee2115660f925c1a0fd3c7dd9b5d9","profile":"core","run_group":"limite-violetto-v2-core-2048","run_id":"8eef4bab-617a-4638-9e51-53e72034e821","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"3"}],"schema_version":"2","sensitivity":{"capabilities":[],"dimensions":[],"models":[],"points":[],"schema_version":"1","sweeps":[]}},"runs":{"16a842b0-5dcc-47b5-8808-850b2ad1ac10":{"cells":[{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:27:15.751735+00:00","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","failure_count":0,"model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.9,"profile":"core","run_id":"16a842b0-5dcc-47b5-8808-850b2ad1ac10","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"16a842b0-5dcc-47b5-8808-850b2ad1ac10","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","family":"limite","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","parameters_b":1.0,"provider_key":"transformers","quantization":null,"run_id":"16a842b0-5dcc-47b5-8808-850b2ad1ac10","runtime_key":"transformers-local"}],"resources":[],"run":{"completed_at_utc":"2026-10-16T08:27:15.751735+00:00","created_at_utc":"2026-10-16T08:17:38.469243+00:00","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","integrity_state":"VALID","manifest_checksum":"89867f6e312b4a47ba55037d176da6aedc3e9ab8e0a8cd12a14605f2c0bddfef","profile":"core","run_group":"limite-mathematics","run_id":"16a842b0-5dcc-47b5-8808-850b2ad1ac10","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"3"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"bf0f533c-e39c-476a-8c5c-e2a2a152276b","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"limite-1b-violetto\"], \"profile\": \"core\", \"run_group\": \"limite-mathematics\", \"run_id\": \"16a842b0-5dcc-47b5-8808-850b2ad1ac10\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"3\"}","model_key":null,"run_id":"16a842b0-5dcc-47b5-8808-850b2ad1ac10","sample_id":null,"status":null,"timestamp_utc":"2026-10-16T08:17:38.469243+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f0a7b650-b682-4663-bcfb-a1aeb0b00117","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"provider_key\": \"transformers\", \"run_group\": \"limite-mathematics\", \"run_id\": \"16a842b0-5dcc-47b5-8808-850b2ad1ac10\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"16a842b0-5dcc-47b5-8808-850b2ad1ac10","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:17:38.489651+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.25412498507648706,"error_message":null,"error_type":null,"event_id":"198fe594-d789-4a7c-aa4e-b2eea47d1ed8","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"provider_key\": \"transformers\", \"run_group\": \"limite-mathematics\", \"run_id\": \"16a842b0-5dcc-47b5-8808-850b2ad1ac10\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"16a842b0-5dcc-47b5-8808-850b2ad1ac10","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:17:38.490392+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"be341e9c-9fc9-436e-8a94-c6154c7a6e3e","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"provider_key\": \"transformers\", \"run_group\": \"limite-mathematics\", \"run_id\": \"16a842b0-5dcc-47b5-8808-850b2ad1ac10\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"16a842b0-5dcc-47b5-8808-850b2ad1ac10","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:17:38.490593+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"cf3e5eae-0ca2-412d-ba86-bc9ffa522e46","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"transformers\", \"run_group\": \"limite-mathematics\", \"run_id\": \"16a842b0-5dcc-47b5-8808-850b2ad1ac10\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"16a842b0-5dcc-47b5-8808-850b2ad1ac10","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:17:38.514776+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"42c100f4-df61-4e75-b22e-cb6d032d57c9","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"profile\": \"core\", \"provider_key\": \"transformers\", \"run_group\": \"limite-mathematics\", \"run_id\": \"16a842b0-5dcc-47b5-8808-850b2ad1ac10\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"16a842b0-5dcc-47b5-8808-850b2ad1ac10","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:27:15.323789+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a18d4943-3d1b-4a94-bb13-9a6b56abc7e4","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"provider_key\": \"transformers\", \"run_group\": \"limite-mathematics\", \"run_id\": \"16a842b0-5dcc-47b5-8808-850b2ad1ac10\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"16a842b0-5dcc-47b5-8808-850b2ad1ac10","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:27:15.324033+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":419.728958979249,"error_message":null,"error_type":null,"event_id":"6b866c88-22f1-42a2-b4ba-f19d58593423","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"provider_key\": \"transformers\", \"run_group\": \"limite-mathematics\", \"run_id\": \"16a842b0-5dcc-47b5-8808-850b2ad1ac10\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"16a842b0-5dcc-47b5-8808-850b2ad1ac10","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:27:15.743971+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"12edc7d6-ebb1-478a-950a-66d3e5e2d5f9","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"run_group\": \"limite-mathematics\", \"run_id\": \"16a842b0-5dcc-47b5-8808-850b2ad1ac10\"}","model_key":"limite-1b-violetto","run_id":"16a842b0-5dcc-47b5-8808-850b2ad1ac10","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:27:15.751476+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"be53ce59-649f-4fbd-94ff-4d5faed9d463","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 40, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 40, \"run_group\": \"limite-mathematics\", \"run_id\": \"16a842b0-5dcc-47b5-8808-850b2ad1ac10\", \"skipped_cases\": 0}","model_key":null,"run_id":"16a842b0-5dcc-47b5-8808-850b2ad1ac10","sample_id":null,"status":null,"timestamp_utc":"2026-10-16T08:27:15.751735+00:00"}]},"16cd6c17-c9de-422d-8583-1a17959bdf6b":{"cells":[{"benchmark_signature":"sha256:benchmark:718e761e4ba3410c289a5bba91a6474827caaa8da30e644b3df925ba4d16fbfc","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T15:04:59.981803+00:00","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","failure_count":0,"model_key":"decisio-qwen3.5-2b-q4km","model_signature":"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.5974025974025974,"profile":"smoke","run_id":"16cd6c17-c9de-422d-8583-1a17959bdf6b","sample_count":77,"status":"COMPLETED","task_id":"intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"16cd6c17-c9de-422d-8583-1a17959bdf6b","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","family":null,"model_id":"unsloth/qwen3.5-2b","model_key":"decisio-qwen3.5-2b-q4km","model_signature":"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95","parameters_b":null,"provider_key":"decisio","quantization":"Q4_K_M","run_id":"16cd6c17-c9de-422d-8583-1a17959bdf6b","runtime_key":"decisio-local"}],"resources":[],"run":{"completed_at_utc":"2026-10-05T15:04:59.981803+00:00","created_at_utc":"2026-10-05T15:03:28.045415+00:00","git_commit":"2e7b64ca38a2db66a9d8b36910d2b2ac5da0c7ce","integrity_state":"VALID","manifest_checksum":"6390323475ebd07e384f1db1b0fa8387774cf197a643733d9162aa0d1898a474","profile":"smoke","run_group":"decisio-classification-20261005_170326","run_id":"16cd6c17-c9de-422d-8583-1a17959bdf6b","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0286d2d8-4714-4a98-8822-510a19ba93d8","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"intent-classification\"], \"model_keys\": [\"decisio-qwen3.5-2b-q4km\"], \"profile\": \"smoke\", \"run_group\": \"decisio-classification-20261005_170326\", \"run_id\": \"16cd6c17-c9de-422d-8583-1a17959bdf6b\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"16cd6c17-c9de-422d-8583-1a17959bdf6b","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T15:03:28.045415+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f1988514-59bd-40fe-be22-85a39fbf29af","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_170326\", \"run_id\": \"16cd6c17-c9de-422d-8583-1a17959bdf6b\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-2b-q4km","run_id":"16cd6c17-c9de-422d-8583-1a17959bdf6b","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T15:03:28.062634+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2203.642167005455,"error_message":null,"error_type":null,"event_id":"974d3f65-9bf0-4d75-bc8c-32a09bc18682","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_170326\", \"run_id\": \"16cd6c17-c9de-422d-8583-1a17959bdf6b\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-2b-q4km","run_id":"16cd6c17-c9de-422d-8583-1a17959bdf6b","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T15:03:30.266771+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0977abce-c352-44b4-8641-deb8a938ad70","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_170326\", \"run_id\": \"16cd6c17-c9de-422d-8583-1a17959bdf6b\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-2b-q4km","run_id":"16cd6c17-c9de-422d-8583-1a17959bdf6b","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T15:03:30.267776+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0ca3ecf7-5f43-48c8-86d2-ab0d9fafecf8","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:718e761e4ba3410c289a5bba91a6474827caaa8da30e644b3df925ba4d16fbfc\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95\", \"planned_cases\": 77, \"profile\": \"smoke\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_170326\", \"run_id\": \"16cd6c17-c9de-422d-8583-1a17959bdf6b\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-2b-q4km","run_id":"16cd6c17-c9de-422d-8583-1a17959bdf6b","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T15:03:30.273439+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0147b00d-2cdd-4147-ae04-c2a800411efb","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:718e761e4ba3410c289a5bba91a6474827caaa8da30e644b3df925ba4d16fbfc\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95\", \"profile\": \"smoke\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_170326\", \"run_id\": \"16cd6c17-c9de-422d-8583-1a17959bdf6b\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-2b-q4km","run_id":"16cd6c17-c9de-422d-8583-1a17959bdf6b","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T15:04:59.976236+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"cdd0056e-e6f0-40fd-a396-fb424e2706d8","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_170326\", \"run_id\": \"16cd6c17-c9de-422d-8583-1a17959bdf6b\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-2b-q4km","run_id":"16cd6c17-c9de-422d-8583-1a17959bdf6b","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T15:04:59.976474+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.008417002391070127,"error_message":null,"error_type":null,"event_id":"4b879172-4967-4264-b8f4-58414596d495","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_170326\", \"run_id\": \"16cd6c17-c9de-422d-8583-1a17959bdf6b\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-2b-q4km","run_id":"16cd6c17-c9de-422d-8583-1a17959bdf6b","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T15:04:59.976649+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"768d69e6-28f5-4a8b-8d1f-ea3b39f7c2ea","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:718e761e4ba3410c289a5bba91a6474827caaa8da30e644b3df925ba4d16fbfc\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95\", \"run_group\": \"decisio-classification-20261005_170326\", \"run_id\": \"16cd6c17-c9de-422d-8583-1a17959bdf6b\"}","model_key":"decisio-qwen3.5-2b-q4km","run_id":"16cd6c17-c9de-422d-8583-1a17959bdf6b","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T15:04:59.981663+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2112cb73-d496-41d4-aede-5a96cd7d03ba","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 77, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 77, \"run_group\": \"decisio-classification-20261005_170326\", \"run_id\": \"16cd6c17-c9de-422d-8583-1a17959bdf6b\", \"skipped_cases\": 0}","model_key":null,"run_id":"16cd6c17-c9de-422d-8583-1a17959bdf6b","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T15:04:59.981803+00:00"}]},"1e48c427-4e75-4ace-a465-2aec7e2751cd":{"cells":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T15:26:12.447493+00:00","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","failure_count":0,"model_key":"decisio-qwen3.5-2b-q4km","model_signature":"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6038961038961039,"profile":"core","run_id":"1e48c427-4e75-4ace-a465-2aec7e2751cd","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"1e48c427-4e75-4ace-a465-2aec7e2751cd","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","family":null,"model_id":"unsloth/qwen3.5-2b","model_key":"decisio-qwen3.5-2b-q4km","model_signature":"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95","parameters_b":null,"provider_key":"decisio","quantization":"Q4_K_M","run_id":"1e48c427-4e75-4ace-a465-2aec7e2751cd","runtime_key":"decisio-local"}],"resources":[],"run":{"completed_at_utc":"2026-10-05T15:26:12.447493+00:00","created_at_utc":"2026-10-05T15:23:10.924758+00:00","git_commit":"2e7b64ca38a2db66a9d8b36910d2b2ac5da0c7ce","integrity_state":"VALID","manifest_checksum":"1c7344f506f0ae91ebfefb0680ca78bf3c41d0388e4a620edf44d6d3644151ca","profile":"core","run_group":"decisio-classification-20261005_172309","run_id":"1e48c427-4e75-4ace-a465-2aec7e2751cd","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7ee346f0-b454-4a3d-b44d-31cf6b23b27f","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"intent-classification\"], \"model_keys\": [\"decisio-qwen3.5-2b-q4km\"], \"profile\": \"core\", \"run_group\": \"decisio-classification-20261005_172309\", \"run_id\": \"1e48c427-4e75-4ace-a465-2aec7e2751cd\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"1e48c427-4e75-4ace-a465-2aec7e2751cd","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T15:23:10.924758+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"87b248f6-d39a-4dd1-a5a5-31ab932af781","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_172309\", \"run_id\": \"1e48c427-4e75-4ace-a465-2aec7e2751cd\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-2b-q4km","run_id":"1e48c427-4e75-4ace-a465-2aec7e2751cd","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T15:23:10.945456+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":3162.3202079936163,"error_message":null,"error_type":null,"event_id":"cb1dd9f0-a753-435a-9ab3-b7bbd8009209","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_172309\", \"run_id\": \"1e48c427-4e75-4ace-a465-2aec7e2751cd\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-2b-q4km","run_id":"1e48c427-4e75-4ace-a465-2aec7e2751cd","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T15:23:14.108490+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e29233df-df1e-4e0b-97b8-48b484dda65c","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_172309\", \"run_id\": \"1e48c427-4e75-4ace-a465-2aec7e2751cd\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-2b-q4km","run_id":"1e48c427-4e75-4ace-a465-2aec7e2751cd","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T15:23:14.109860+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"53e94ac7-4808-4561-8415-c128fcf97858","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_172309\", \"run_id\": \"1e48c427-4e75-4ace-a465-2aec7e2751cd\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-2b-q4km","run_id":"1e48c427-4e75-4ace-a465-2aec7e2751cd","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T15:23:14.118476+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d5a93d4b-e895-4492-bd5e-e7d43d92f9cb","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95\", \"profile\": \"core\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_172309\", \"run_id\": \"1e48c427-4e75-4ace-a465-2aec7e2751cd\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-2b-q4km","run_id":"1e48c427-4e75-4ace-a465-2aec7e2751cd","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T15:26:12.439068+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d5ad303b-7911-4a9b-a1d4-8bc7fe09d4da","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_172309\", \"run_id\": \"1e48c427-4e75-4ace-a465-2aec7e2751cd\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-2b-q4km","run_id":"1e48c427-4e75-4ace-a465-2aec7e2751cd","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T15:26:12.439259+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.013083001249469817,"error_message":null,"error_type":null,"event_id":"162f7331-e125-46f9-8fd7-ffbc08317835","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_172309\", \"run_id\": \"1e48c427-4e75-4ace-a465-2aec7e2751cd\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-2b-q4km","run_id":"1e48c427-4e75-4ace-a465-2aec7e2751cd","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T15:26:12.439390+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"4f0ea60f-1d21-4347-b828-d45715eac36a","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:82f892aa7dd27a1e40624b83df8f861d41b03bf6c2f1a3f523b7804ca8998c95\", \"run_group\": \"decisio-classification-20261005_172309\", \"run_id\": \"1e48c427-4e75-4ace-a465-2aec7e2751cd\"}","model_key":"decisio-qwen3.5-2b-q4km","run_id":"1e48c427-4e75-4ace-a465-2aec7e2751cd","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T15:26:12.447294+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"5d0abdc2-a698-44e3-8df3-2df2212c06e0","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"decisio-classification-20261005_172309\", \"run_id\": \"1e48c427-4e75-4ace-a465-2aec7e2751cd\", \"skipped_cases\": 0}","model_key":null,"run_id":"1e48c427-4e75-4ace-a465-2aec7e2751cd","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T15:26:12.447493+00:00"}]},"2026-10-core-v1-bonsai-27b-q10-intent-classification-core":{"cells":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-06T10:03:19.101787+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.5974025974025974,"profile":"core","run_id":"2026-10-core-v1-bonsai-27b-q10-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-bonsai-27b-q10-intent-classification-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"bonsai-27b-q10","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","parameters_b":null,"provider_key":"korgis","quantization":"Q1_0","run_id":"2026-10-core-v1-bonsai-27b-q10-intent-classification-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"bonsai-27b-q10","process_cpu_percent_avg":4.141997751956015,"process_rss_bytes_avg":5899303355.703257,"process_rss_bytes_peak":10491068416.0,"sample_count":4416,"sample_interval_ms":317.0853588508965,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4949098496.0}],"run":{"completed_at_utc":"2026-10-06T10:03:19.101787+00:00","created_at_utc":"2026-10-06T09:40:30.604010+00:00","git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","integrity_state":"VALID","manifest_checksum":"6dbc1d35e1ad6b5e63a85e9eb837b3eca7648b3789e4935e901d82f096173382","profile":"core","run_group":"2026-10-core-v1-bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-intent-classification-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"792a9a66-c2e6-42d5-a021-564a4d7dbae2","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"intent-classification\"], \"model_keys\": [\"bonsai-27b-q10\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-intent-classification-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-bonsai-27b-q10-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T09:40:30.604010+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b4dfd62f-1c79-4a17-a9d5-435172bc9599","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:40:30.635226+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":3118.7333749985555,"error_message":null,"error_type":null,"event_id":"99e1ef9e-5f07-443a-abf3-f596e1fa119a","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:40:33.754914+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3aee6ea0-844d-4ed2-8f49-dd3ef6e67cdf","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:40:33.755462+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"989c0b71-79fc-4c10-ad32-0dca7e8150fe","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:40:33.769302+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ff18f138-ebe1-4afa-8fe8-5225093db599","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T10:03:18.511160+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8fcbbc9e-da7c-43ea-9376-590985ecd241","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T10:03:18.511662+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":571.1732500058133,"error_message":null,"error_type":null,"event_id":"4a847c1f-0281-4748-bdb7-00b98ce3d53b","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T10:03:19.083002+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"658bb430-f581-4a44-a525-97b16a477276","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-intent-classification-core\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T10:03:19.101625+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ba827175-6823-4bd1-8bd2-d4fd51a6134f","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-intent-classification-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-bonsai-27b-q10-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T10:03:19.101787+00:00"}]},"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core":{"cells":[{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-06T09:40:29.476460+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.575,"profile":"core","run_id":"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"bonsai-27b-q10","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","parameters_b":null,"provider_key":"korgis","quantization":"Q1_0","run_id":"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"bonsai-27b-q10","process_cpu_percent_avg":10.398893667073832,"process_rss_bytes_avg":7636308908.292063,"process_rss_bytes_peak":11685068800.0,"sample_count":340,"sample_interval_ms":307.1656581635226,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5910167552.0}],"run":{"completed_at_utc":"2026-10-06T09:40:29.476460+00:00","created_at_utc":"2026-10-06T09:38:50.359790+00:00","git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","integrity_state":"VALID","manifest_checksum":"cf36c409d357a074bd5ff5a9872047e7cb53d64b51a91a9ede5f0a5be30dc332","profile":"core","run_group":"2026-10-core-v1-bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7fdc274d-73ed-49f5-b041-4ab93f47f07b","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"model_keys\": [\"bonsai-27b-q10\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T09:38:50.359790+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"de43ed13-da2f-4e0f-81a9-927542e4697f","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:38:50.392440+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":3112.0705420034938,"error_message":null,"error_type":null,"event_id":"f85f2a18-0863-421b-8259-bd17e4f5cf82","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:38:53.505244+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"29e4d119-222e-4d29-9785-abc98e9dc8ca","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:38:53.505602+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8f7375bd-307b-46f1-8791-7a6680904719","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:38:53.521244+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6f13d91f-0229-48f8-92b2-e5ca922107df","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:40:28.859204+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"35bb19ea-297d-4f5d-90f3-3acdda457477","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:40:28.859410+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":611.1365840042708,"error_message":null,"error_type":null,"event_id":"403cdd8f-5b14-4e8a-8308-cc5af8d24023","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:40:29.470729+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8078abaf-671f-4f73-b799-d68cdef28603","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:40:29.476185+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d39a49d5-8de8-4033-b782-5d2a825e9cd6","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 40, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 40, \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-bonsai-27b-q10-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T09:40:29.476460+00:00"}]},"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core":{"cells":[{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-06T10:26:04.512560+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.948051948051948,"profile":"core","run_id":"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"bonsai-27b-q10","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","parameters_b":null,"provider_key":"korgis","quantization":"Q1_0","run_id":"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"bonsai-27b-q10","process_cpu_percent_avg":3.871693631322875,"process_rss_bytes_avg":5867857169.246841,"process_rss_bytes_peak":9145516032.0,"sample_count":4524,"sample_interval_ms":309.6376288065501,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5604933632.0}],"run":{"completed_at_utc":"2026-10-06T10:26:04.512560+00:00","created_at_utc":"2026-10-06T10:03:19.887857+00:00","git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","integrity_state":"VALID","manifest_checksum":"9bccd97bd698de1be4710f3e55169bb9060f349b79417cb2a40f74b2f9f795cb","profile":"core","run_group":"2026-10-core-v1-bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c3fdf1bd-e169-4029-b3ac-fd9b1c27a486","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"oos-calibration\"], \"model_keys\": [\"bonsai-27b-q10\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T10:03:19.887857+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"afcb475a-3c6b-44d4-8d2a-6b189a479d31","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T10:03:19.903868+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2098.8719159940956,"error_message":null,"error_type":null,"event_id":"4558d1e5-fcf7-43ec-8e7d-5fe7564a25c4","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T10:03:22.003146+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6949a4c0-0f22-4f16-9674-82768b2b4e84","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T10:03:22.003785+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"91e64b55-c22b-4938-b146-9859786b4fc6","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T10:03:22.027758+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"921245cf-38c7-4046-b344-171a41c30f3a","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T10:26:03.959595+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d9908af2-c32c-414a-926c-7d3a58658296","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T10:26:03.961057+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":523.0728330061538,"error_message":null,"error_type":null,"event_id":"27b9cbe6-c475-42bd-b291-26e87d323e47","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T10:26:04.484476+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"975a7138-dfca-400c-a793-9bfee655e4f3","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T10:26:04.512386+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"efe80248-47bb-41c5-a842-260fe7845665","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-bonsai-27b-q10-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T10:26:04.512560+00:00"}]},"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core":{"cells":[{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-06T09:38:49.212082+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"profile":"core","run_id":"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"bonsai-27b-q10","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","parameters_b":null,"provider_key":"korgis","quantization":"Q1_0","run_id":"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"bonsai-27b-q10","process_cpu_percent_avg":9.839476147142088,"process_rss_bytes_avg":6890131131.833357,"process_rss_bytes_peak":10222600192.0,"sample_count":690,"sample_interval_ms":311.07264410993747,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5791531008.0}],"run":{"completed_at_utc":"2026-10-06T09:38:49.212082+00:00","created_at_utc":"2026-10-06T09:35:24.100477+00:00","git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","integrity_state":"VALID","manifest_checksum":"6c08aa699a3a5650d852c55c805675ce41ffc70f36278f90318d3538c54b1136","profile":"core","run_group":"2026-10-core-v1-bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"936546ae-eaa1-418a-95ec-ab6a3938d68a","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"qa-abstention\"], \"model_keys\": [\"bonsai-27b-q10\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T09:35:24.100477+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1d9b4d42-bdb0-4723-a9a2-36e8c0332261","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:35:24.130565+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":3118.1472500029486,"error_message":null,"error_type":null,"event_id":"ad346ce2-038d-4031-b8b8-e3cdc3ed6148","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:35:27.249425+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0a238c21-88d0-4d1e-89b5-cae15c4d6c46","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:35:27.249866+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"92ab02b8-3e45-4bf1-aaa6-b773ef4bedf2","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:35:27.282854+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f1dc8982-2ae9-46ee-9205-9094aa0f08f5","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:38:48.484265+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a2dd1330-a9a0-4dc8-8fd6-ef1b628161a7","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:38:48.484487+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":719.1172089951579,"error_message":null,"error_type":null,"event_id":"ab19c08b-6148-4fea-bdd6-b4dcc18a453b","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:38:49.203773+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"fbbe57e7-f709-4a5f-b33e-9ac3ad4547b9","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:38:49.211858+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3fe195ee-a647-4a3e-83d1-bc7676b3e6ff","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-bonsai-27b-q10-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T09:38:49.212082+00:00"}]},"2026-10-core-v1-bonsai-27b-q10-structured-output-core":{"cells":[{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-06T09:35:22.771531+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7833333333333333,"profile":"core","run_id":"2026-10-core-v1-bonsai-27b-q10-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-bonsai-27b-q10-structured-output-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"bonsai-27b-q10","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","parameters_b":null,"provider_key":"korgis","quantization":"Q1_0","run_id":"2026-10-core-v1-bonsai-27b-q10-structured-output-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"bonsai-27b-q10","process_cpu_percent_avg":9.52518999663954,"process_rss_bytes_avg":6054163284.918202,"process_rss_bytes_peak":8022573056.0,"sample_count":990,"sample_interval_ms":317.9005326878062,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5717721088.0}],"run":{"completed_at_utc":"2026-10-06T09:35:22.771531+00:00","created_at_utc":"2026-10-06T09:30:16.582279+00:00","git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","integrity_state":"VALID","manifest_checksum":"6f4c1fd7b965df252b28d3d973043e7221babe98aadd5097fc35e49daf990d91","profile":"core","run_group":"2026-10-core-v1-bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-structured-output-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"bccad7ab-5e57-4e46-95ee-4968e6f4f03e","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"structured-output\"], \"model_keys\": [\"bonsai-27b-q10\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-structured-output-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-bonsai-27b-q10-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T09:30:16.582279+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"df6cbd14-027d-4cec-bcd1-652db0d33a73","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:30:16.608904+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":3107.1267499937676,"error_message":null,"error_type":null,"event_id":"9b6579c6-f23f-40f1-b1da-f4f28296d1b7","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:30:19.716801+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a27a6c4d-21eb-4daf-98c1-1206a04562c1","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:30:19.717223+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2866a3f6-b582-43e7-8082-0a6d4ec139ba","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:30:19.793525+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7221aa91-b533-420c-a45d-df6a7e9a7cb9","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:35:21.996662+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1ef167e7-7e94-40d6-bd36-3c2f6d1317c7","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:35:21.997177+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":766.8199160107179,"error_message":null,"error_type":null,"event_id":"a617a517-6e18-49f6-af85-2db5c1cd800c","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:35:22.764224+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"939f37a8-e264-42e6-9be6-0113a18616e4","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-structured-output-core\"}","model_key":"bonsai-27b-q10","run_id":"2026-10-core-v1-bonsai-27b-q10-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:35:22.771271+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f91c09dd-533c-43f9-b01d-09b80627c220","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"2026-10-core-v1-bonsai-27b-q10\", \"run_id\": \"2026-10-core-v1-bonsai-27b-q10-structured-output-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-bonsai-27b-q10-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T09:35:22.771531+00:00"}]},"2026-10-core-v1-gpt-5.6-luna-intent-classification-core":{"cells":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-06T12:22:05.300981+00:00","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","failure_count":0,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.8246753246753247,"profile":"core","run_id":"2026-10-core-v1-gpt-5.6-luna-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-intent-classification-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"gpt-5.6-luna","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","family":null,"model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","parameters_b":null,"provider_key":"openai","quantization":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-intent-classification-core","runtime_key":"openai-api"}],"resources":[],"run":{"completed_at_utc":"2026-10-06T12:22:05.300981+00:00","created_at_utc":"2026-10-06T12:17:37.350706+00:00","git_commit":"11b00f55c776ec89ddee7cd2dd1898797dde4055","integrity_state":"VALID","manifest_checksum":"bd7c5aea73cc65daf8f7831b00da10ab6bb0a50e70988a1c5adb0e6f4085aac7","profile":"core","run_group":"2026-10-core-v1-gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-intent-classification-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8799b328-3d51-49b5-b0fe-c9d05b3ad5aa","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"intent-classification\"], \"model_keys\": [\"gpt-5.6-luna\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-intent-classification-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T12:17:37.350706+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f26cb063-4071-4e88-9ed2-a5b7d675edbd","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-intent-classification-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:17:37.369515+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":35.03787500085309,"error_message":null,"error_type":null,"event_id":"0a638a73-d614-4003-9e0c-1a784935042f","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-intent-classification-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:17:37.405038+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ceb1918f-c07f-4884-8b2e-75f184225dfb","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-intent-classification-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:17:37.405282+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3ced420c-c74b-40b4-94ab-471fb5423033","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-intent-classification-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:17:37.410625+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7097e54a-a302-4496-9cb1-5d27ab0ff1db","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"profile\": \"core\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-intent-classification-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:22:05.288983+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2f570093-5448-43c1-883c-e22bf336daab","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-intent-classification-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:22:05.289201+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.01054200401995331,"error_message":null,"error_type":null,"event_id":"de34f82f-ebe4-4c54-ab61-be2ec2914a8d","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-intent-classification-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:22:05.289368+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"9cc4fb74-a61e-4aeb-adcb-fd704e234d97","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-intent-classification-core\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:22:05.300796+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"925227ad-1a22-4c6a-bfee-bda3b639dedb","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-intent-classification-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T12:22:05.300981+00:00"}]},"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core":{"cells":[{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-06T12:17:36.689339+00:00","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","failure_count":0,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.45,"profile":"core","run_id":"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"gpt-5.6-luna","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","family":null,"model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","parameters_b":null,"provider_key":"openai","quantization":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core","runtime_key":"openai-api"}],"resources":[],"run":{"completed_at_utc":"2026-10-06T12:17:36.689339+00:00","created_at_utc":"2026-10-06T12:16:27.098212+00:00","git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","integrity_state":"VALID","manifest_checksum":"8ba7c50c2c7736dfc7481a72224fc6bcb09fbd400ed74b19a413baaa109b19b5","profile":"core","run_group":"2026-10-core-v1-gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"143b5c0f-b38d-4808-beec-0436d5bdb261","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"model_keys\": [\"gpt-5.6-luna\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T12:16:27.098212+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"cadce18d-d064-46da-bb88-868febb64844","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:16:27.118444+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":36.60933299397584,"error_message":null,"error_type":null,"event_id":"5259893f-a800-4a89-a542-88df49fa4e19","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:16:27.155551+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d5c8372b-9979-41e1-a51b-552df26f5042","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:16:27.156441+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d5abc3da-6896-4c54-818e-d00e24033abd","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:16:27.167284+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8041a0fa-b47c-455a-b3ac-d9368c1bac75","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"profile\": \"core\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:17:36.672812+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f8e197ea-a850-407c-8504-f1b090c62d6b","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:17:36.673185+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.029709000955335796,"error_message":null,"error_type":null,"event_id":"b81eb321-92fe-4bb8-b3eb-e8d3a7e8caf9","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:17:36.673639+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"674361f1-87ad-40a0-a6ea-88d2017aade6","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:17:36.689149+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7dabde50-6568-4f39-bc80-0d29980a5794","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 40, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 40, \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T12:17:36.689339+00:00"}]},"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core":{"cells":[{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-06T12:26:04.935243+00:00","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","failure_count":0,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":1.0,"profile":"core","run_id":"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"gpt-5.6-luna","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","family":null,"model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","parameters_b":null,"provider_key":"openai","quantization":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core","runtime_key":"openai-api"}],"resources":[],"run":{"completed_at_utc":"2026-10-06T12:26:04.935243+00:00","created_at_utc":"2026-10-06T12:22:06.019355+00:00","git_commit":"11b00f55c776ec89ddee7cd2dd1898797dde4055","integrity_state":"VALID","manifest_checksum":"19ec62d9277d6b6b531361db8fce74522d4f499abc2ce1bbd308d40fa9ff0a8c","profile":"core","run_group":"2026-10-core-v1-gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7c24aef3-84af-4c7d-af90-f604cac0fd32","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"oos-calibration\"], \"model_keys\": [\"gpt-5.6-luna\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T12:22:06.019355+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b239cdd8-f213-41ef-807f-e56da38ea62e","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:22:06.038143+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":36.37604200048372,"error_message":null,"error_type":null,"event_id":"b3e00642-b89b-42b8-a97f-367ae5bfb4e9","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:22:06.074943+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b3e5199f-0e50-4aa2-a387-210b8e27cb74","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:22:06.075219+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8b422b9e-239e-44ca-9139-b8b04e5889cb","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:22:06.094753+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a13f63ae-74d4-446c-a70b-764b6920aaa3","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"profile\": \"core\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:26:04.910780+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"93e9c444-1839-4b57-a6d5-b8f65097e0e8","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:26:04.911227+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.021375002688728273,"error_message":null,"error_type":null,"event_id":"f0fbe27a-b136-4de0-8793-2bbcbd020b85","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:26:04.911550+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"19b7c709-e130-4329-b47b-cd94a6c64ef2","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:26:04.935014+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"79379af5-4064-4958-b536-6aeb65a4180d","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T12:26:04.935243+00:00"}]},"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core":{"cells":[{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-06T12:16:26.396465+00:00","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","failure_count":0,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"profile":"core","run_id":"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"gpt-5.6-luna","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","family":null,"model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","parameters_b":null,"provider_key":"openai","quantization":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core","runtime_key":"openai-api"}],"resources":[],"run":{"completed_at_utc":"2026-10-06T12:16:26.396465+00:00","created_at_utc":"2026-10-06T12:15:05.482219+00:00","git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","integrity_state":"VALID","manifest_checksum":"6c0f05e4061c6bd973da3239c2bde22948606c739405ad4017cc8de1de0b4cc1","profile":"core","run_group":"2026-10-core-v1-gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8e927750-2ca3-45e7-908f-d82291d94c18","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"qa-abstention\"], \"model_keys\": [\"gpt-5.6-luna\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T12:15:05.482219+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"29ed5bc1-8aaf-4c63-92d6-4917b3f06f4a","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:15:05.504149+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":38.12149999430403,"error_message":null,"error_type":null,"event_id":"aeb06635-0d76-446a-993f-80dfc7f91c83","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:15:05.542659+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"725212c9-be40-4974-a4a4-b8182ae0ae42","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:15:05.542904+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a144bb00-a22d-485f-bb12-0ba4a7af5f63","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:15:05.563111+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c094598a-87ec-4e85-b6cd-743b0a09706e","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"profile\": \"core\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:16:26.385956+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8bc75acc-7a5f-4984-b68a-95aab2de4ec4","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:16:26.386278+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.01620801049284637,"error_message":null,"error_type":null,"event_id":"0f1af03d-4d8e-40ca-bf87-d0e79c36b6a9","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:16:26.386575+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"08de29a8-cf8b-4bee-95a8-ea6bcc4aded2","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:16:26.396205+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"53034a81-c5bf-4e93-a44a-d4c829a2b04d","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T12:16:26.396465+00:00"}]},"2026-10-core-v1-gpt-5.6-luna-structured-output-core":{"cells":[{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-06T12:15:04.761551+00:00","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","failure_count":0,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.9166666666666666,"profile":"core","run_id":"2026-10-core-v1-gpt-5.6-luna-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-structured-output-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"gpt-5.6-luna","execution_signature":"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3","family":null,"model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","parameters_b":null,"provider_key":"openai","quantization":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-structured-output-core","runtime_key":"openai-api"}],"resources":[],"run":{"completed_at_utc":"2026-10-06T12:15:04.761551+00:00","created_at_utc":"2026-10-06T12:13:07.662669+00:00","git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","integrity_state":"VALID","manifest_checksum":"86a0e4a754679fb7152e944bb5539726a81773c6fc042bf83cbf0afefce68bac","profile":"core","run_group":"2026-10-core-v1-gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-structured-output-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1d22bd5d-da8e-458d-ae71-ffd456cdf2cc","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"structured-output\"], \"model_keys\": [\"gpt-5.6-luna\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-structured-output-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T12:13:07.662669+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0d312057-300c-4049-8a05-3a7a494ab212","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-structured-output-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:13:07.684826+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":36.28937499888707,"error_message":null,"error_type":null,"event_id":"cca91a24-ed48-48e7-856e-d8ff99961212","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-structured-output-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:13:07.721859+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"71c995f6-7fbe-478c-8454-c0880247681f","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-structured-output-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:13:07.722077+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"28896f1d-c669-445e-9212-18f54214013e","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-structured-output-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:13:07.769147+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ce6c70cd-aa3a-4be1-b076-4547480f3b9f","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"profile\": \"core\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-structured-output-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:15:04.749178+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"bfba5c19-e973-4466-a084-1f85f6b3ae01","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-structured-output-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:15:04.749860+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.025957997422665358,"error_message":null,"error_type":null,"event_id":"7ccd838a-af70-434d-b99f-a1726b85a839","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-structured-output-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:15:04.750401+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"72a4eb61-8c86-4364-87e2-3e7ac6426b67","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:1a6c616519430dc14507dd625d0575ae7bddea41f1861c284a266994ee519ac3\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-structured-output-core\"}","model_key":"gpt-5.6-luna","run_id":"2026-10-core-v1-gpt-5.6-luna-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:15:04.761266+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f226b407-1b69-4f78-86b1-04b5fbb5e30c","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"2026-10-core-v1-gpt-5.6-luna\", \"run_id\": \"2026-10-core-v1-gpt-5.6-luna-structured-output-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-gpt-5.6-luna-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T12:15:04.761551+00:00"}]},"2026-10-core-v1-jev-1.13.0-intent-classification-core":{"cells":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-06T12:50:59.652242+00:00","execution_signature":"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3","failure_count":0,"model_key":"jev-1.13.0","model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.7467532467532467,"profile":"core","run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"jev-1.13.0","execution_signature":"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3","family":"jev","model_id":"jev-1.13.0","model_key":"jev-1.13.0","model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","parameters_b":null,"provider_key":"typesafe","quantization":null,"run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","runtime_key":"typesafe-api"}],"resources":[],"run":{"completed_at_utc":"2026-10-06T12:50:59.652242+00:00","created_at_utc":"2026-10-05T14:40:39.618610+00:00","git_commit":"72697127df45f8a3abc98b0b7bdf6fd53a0b5c98","integrity_state":"VALID","manifest_checksum":"de56ecd6fb13c6af2e0f2fd5e4e5465e3ef0b3a99b8d92489d91de51962ffa9a","profile":"core","run_group":"2026-10-core-v1-jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ae162d37-2ca9-4e5f-9c49-87ef4d0194b9","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"intent-classification\"], \"model_keys\": [\"jev-1.13.0\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-intent-classification-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T14:40:39.618610+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"488410e1-0b4c-450d-b88f-72490912aa7f","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:22b2f9a66916523e6400c22d60c650411dde9295e3009f1e0c128177f9752535\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-intent-classification-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T14:40:39.634445+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":66.94120800239034,"error_message":null,"error_type":null,"event_id":"2f25ba11-09d7-4858-9750-8e0524f155c7","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:22b2f9a66916523e6400c22d60c650411dde9295e3009f1e0c128177f9752535\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-intent-classification-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T14:40:39.701825+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"4d0edf30-dc7e-4bdf-9c66-1a7a8c6acb1a","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:22b2f9a66916523e6400c22d60c650411dde9295e3009f1e0c128177f9752535\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-intent-classification-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T14:40:39.702079+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c5bb6d16-7243-44fa-aa9e-446b650ffc66","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:22b2f9a66916523e6400c22d60c650411dde9295e3009f1e0c128177f9752535\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-intent-classification-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T14:40:39.706460+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c9d4135b-48b3-4828-8f3b-b8a4d4d7e8f0","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:22b2f9a66916523e6400c22d60c650411dde9295e3009f1e0c128177f9752535\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"profile\": \"core\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-intent-classification-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T14:41:20.724296+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a57b41be-a062-40f9-9352-0e01d72a5cb3","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:22b2f9a66916523e6400c22d60c650411dde9295e3009f1e0c128177f9752535\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-intent-classification-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T14:41:20.724487+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.0015420082490891218,"error_message":null,"error_type":null,"event_id":"371b4b4f-c0cd-4151-b8b4-8dac9e0661e8","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:22b2f9a66916523e6400c22d60c650411dde9295e3009f1e0c128177f9752535\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-intent-classification-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T14:41:20.724618+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a978b64b-7f42-413f-8708-691c0a36aa23","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:22b2f9a66916523e6400c22d60c650411dde9295e3009f1e0c128177f9752535\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-intent-classification-core\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T14:41:20.734937+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"14b2260a-04c9-41c4-912a-e22870fed0e0","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-intent-classification-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T14:41:20.735303+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"35823a47-682e-46eb-b635-c09a8e36290f","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"intent-classification\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"jev-1.13.0\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-intent-classification-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T12:50:19.453021+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"74869886-a52f-434c-8ed1-6a7db831bcf2","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-intent-classification-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:50:19.473735+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":54.8362499976065,"error_message":null,"error_type":null,"event_id":"dcd418cb-5d6a-4b07-aaa3-11a20c45ee69","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-intent-classification-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:50:19.528921+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6995a054-e12d-429c-a0ad-b743e72cd272","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-intent-classification-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:50:19.529112+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"bae643a9-e638-4271-8991-46c2c19c2cf1","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-intent-classification-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:50:19.532949+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2ab101b2-d50e-4d6a-9101-a3ee1e0cbbcf","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"profile\": \"core\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-intent-classification-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:50:59.620976+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f6e17b3a-875a-448f-bc3b-e3d2b3314aae","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-intent-classification-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:50:59.621360+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.026125009753741324,"error_message":null,"error_type":null,"event_id":"c8bd61bf-d312-4a84-b81f-8b92bf7c401c","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-intent-classification-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:50:59.621731+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"9fd1f619-3ffc-4d91-ab11-09f3a804e004","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-intent-classification-core\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:50:59.652028+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6c0695c5-7ad4-45a8-8721-778a941ff85e","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-intent-classification-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-jev-1.13.0-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T12:50:59.652242+00:00"}]},"2026-10-core-v1-jev-1.13.0-oos-calibration-core":{"cells":[{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-06T12:51:50.754874+00:00","execution_signature":"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3","failure_count":0,"model_key":"jev-1.13.0","model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.974025974025974,"profile":"core","run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"jev-1.13.0","execution_signature":"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3","family":"jev","model_id":"jev-1.13.0","model_key":"jev-1.13.0","model_signature":"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6","parameters_b":null,"provider_key":"typesafe","quantization":null,"run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","runtime_key":"typesafe-api"}],"resources":[],"run":{"completed_at_utc":"2026-10-06T12:51:50.754874+00:00","created_at_utc":"2026-10-05T14:41:27.534659+00:00","git_commit":"72697127df45f8a3abc98b0b7bdf6fd53a0b5c98","integrity_state":"VALID","manifest_checksum":"65d7fb3daaaa865e3f603e4ab6cafb98acf0876bf3efb7f37c1fba7796a30c0f","profile":"core","run_group":"2026-10-core-v1-jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e0cab2fd-b339-4a4e-956f-aa95dcaa1860","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"oos-calibration\"], \"model_keys\": [\"jev-1.13.0\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-oos-calibration-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T14:41:27.534659+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"5a928d26-c0d8-4a16-885f-5233175d701a","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:22b2f9a66916523e6400c22d60c650411dde9295e3009f1e0c128177f9752535\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-oos-calibration-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T14:41:27.553957+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":71.80012499156874,"error_message":null,"error_type":null,"event_id":"42f9212d-e647-481e-a6a3-a523808f7a01","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:22b2f9a66916523e6400c22d60c650411dde9295e3009f1e0c128177f9752535\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-oos-calibration-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T14:41:27.626253+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"cee86534-0535-4af2-90ac-e763c1a568fd","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:22b2f9a66916523e6400c22d60c650411dde9295e3009f1e0c128177f9752535\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-oos-calibration-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T14:41:27.626490+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8f477312-969f-4bda-944e-d175b9bc7b44","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:22b2f9a66916523e6400c22d60c650411dde9295e3009f1e0c128177f9752535\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-oos-calibration-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T14:41:27.645074+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"40ec0145-6205-4fcc-b301-efefe96003d2","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:22b2f9a66916523e6400c22d60c650411dde9295e3009f1e0c128177f9752535\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"profile\": \"core\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-oos-calibration-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T14:42:09.749766+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"cf1d9a4e-33c2-40a0-875e-b1d3aceb2ac7","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:22b2f9a66916523e6400c22d60c650411dde9295e3009f1e0c128177f9752535\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-oos-calibration-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T14:42:09.750100+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.0012499949662014842,"error_message":null,"error_type":null,"event_id":"079b637c-b7b6-44bd-be4f-649060a96137","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:22b2f9a66916523e6400c22d60c650411dde9295e3009f1e0c128177f9752535\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-oos-calibration-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T14:42:09.750289+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"4bc6ce41-5c08-48cf-a911-ceb32a2b27c3","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:22b2f9a66916523e6400c22d60c650411dde9295e3009f1e0c128177f9752535\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-oos-calibration-core\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T14:42:09.760641+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"58c55392-de88-45ef-8fd2-418866403c78","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-oos-calibration-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T14:42:09.760841+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"50187160-5bc3-445e-ad90-54198b9ad508","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"oos-calibration\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"jev-1.13.0\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-oos-calibration-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T12:51:09.271508+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"425ec944-ec78-49b3-8253-6014d4db2b5f","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-oos-calibration-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:51:09.289723+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":62.519958009943366,"error_message":null,"error_type":null,"event_id":"1f95b4c3-c547-47b6-be32-027751403ef3","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-oos-calibration-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:51:09.352645+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"9ad71d99-cf12-4219-bb6b-1f64ee07514c","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-oos-calibration-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:51:09.352861+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"06e73a43-927c-4ccb-87e6-0462bd62bf43","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-oos-calibration-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:51:09.374922+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1159fd5b-fa53-4131-b69d-cf124cfbf451","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"profile\": \"core\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-oos-calibration-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:51:50.718511+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"36b7749b-9905-4b84-9bbf-f86a67a0cf37","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-oos-calibration-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T12:51:50.719124+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.09420800779480487,"error_message":null,"error_type":null,"event_id":"57c899a5-cdf3-4794-9658-098c1f12bb64","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"provider_key\": \"typesafe\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-oos-calibration-core\", \"runtime_key\": \"typesafe-api\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:51:50.719778+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f803941a-96c0-4144-a60a-550d4768436d","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:78d651f162c10df45c1c046427893fb4c92ba19b65bc80e4c02e0901408e1ce3\", \"model_key\": \"jev-1.13.0\", \"model_signature\": \"sha256:model:132b58cbd2443d5095b774bef2b7574b957711e27d1fc5048133acb5364acae6\", \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-oos-calibration-core\"}","model_key":"jev-1.13.0","run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T12:51:50.754572+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"be8efec7-0efb-4c4b-93ec-17fd47d8abda","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"2026-10-core-v1-jev-1.13.0\", \"run_id\": \"2026-10-core-v1-jev-1.13.0-oos-calibration-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-jev-1.13.0-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T12:51:50.754874+00:00"}]},"2026-10-core-v1-nemotron-nano-4b-intent-classification-core":{"cells":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T12:03:20.624151+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.5584415584415584,"profile":"core","run_id":"2026-10-core-v1-nemotron-nano-4b-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-intent-classification-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"nemotron-nano-4b","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","parameters_b":null,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"2026-10-core-v1-nemotron-nano-4b-intent-classification-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b","process_cpu_percent_avg":5.91751681481258,"process_rss_bytes_avg":5777497087.999999,"process_rss_bytes_peak":10717691904.0,"sample_count":1252,"sample_interval_ms":293.3800480326384,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6120767488.0}],"run":{"completed_at_utc":"2026-10-05T12:03:20.624151+00:00","created_at_utc":"2026-10-05T11:57:48.020116+00:00","git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","integrity_state":"VALID","manifest_checksum":"97d5314d35237c352b5707b3257a5733850187d31cbfc94175a79545c72660af","profile":"core","run_group":"2026-10-core-v1-nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-intent-classification-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f249ba9a-162a-44ac-945d-f725051898c8","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"intent-classification\"], \"model_keys\": [\"nemotron-nano-4b\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-intent-classification-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T11:57:48.020116+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2ee4bc9d-8b90-4b3d-acb5-55012a54cd96","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:57:48.041318+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":1088.7861660012277,"error_message":null,"error_type":null,"event_id":"9529cace-473f-4050-8de6-6c3253403dda","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:57:49.130706+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"406fc62b-9342-4e07-bd95-14311a5750a6","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:57:49.131144+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"de9dea50-bc43-4d04-a1c0-bc0855fa2ad6","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:57:49.139202+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c24829aa-2294-4d3c-9efc-1fbab9e224d8","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T12:03:20.226210+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2d4a2258-51e4-4926-a4d2-312c1e7332d4","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T12:03:20.226369+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":387.6089170080377,"error_message":null,"error_type":null,"event_id":"453c8141-22b4-45f8-8dc1-2bded2e81d4e","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T12:03:20.614162+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"18dfa316-dbe6-4f29-b4f9-e56f220fb7b2","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-intent-classification-core\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T12:03:20.623967+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7e031879-7078-4e48-a17a-152f76e9dce9","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-intent-classification-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T12:03:20.624151+00:00"}]},"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core":{"cells":[{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T11:57:43.663487+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.575,"profile":"core","run_id":"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"nemotron-nano-4b","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","parameters_b":null,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b","process_cpu_percent_avg":11.038100201393934,"process_rss_bytes_avg":7133634082.133334,"process_rss_bytes_peak":11070996480.0,"sample_count":143,"sample_interval_ms":255.84691734608595,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":11042144256.0}],"run":{"completed_at_utc":"2026-10-05T11:57:43.663487+00:00","created_at_utc":"2026-10-05T11:57:14.293645+00:00","git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","integrity_state":"VALID","manifest_checksum":"b35a42b1108e1d34bb4b0eb737aa8238a0524800b4318c1d29a5c6a29479817f","profile":"core","run_group":"2026-10-core-v1-nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f48b3985-a79a-4abe-90f6-48655de9066f","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"model_keys\": [\"nemotron-nano-4b\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T11:57:14.293645+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f416072f-b937-47b5-a4bd-af2caeffa129","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:57:14.309754+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":1091.060082995682,"error_message":null,"error_type":null,"event_id":"a9fb0c8e-0535-4ade-bade-66e7eef9b74e","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:57:15.401255+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"177e863e-1cbe-4e41-b698-7281340fdbbd","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:57:15.401586+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a37bb58d-c720-4c40-85b2-aa58907ba26d","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:57:15.413068+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"56131e9f-5ce2-4c9d-b0d2-496d008ce95d","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:57:43.482679+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"08438193-3194-4a29-a3a2-228079932ecf","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:57:43.482914+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":176.36108399892692,"error_message":null,"error_type":null,"event_id":"8990c08a-c187-423c-823b-42912e2767c0","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:57:43.659485+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"926330a8-a618-4ef1-aadb-73b97a749f6a","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:57:43.663323+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"4ced8d29-4d23-4680-bef1-7922790d7ee7","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 40, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 40, \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T11:57:43.663487+00:00"}]},"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core":{"cells":[{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T12:10:00.371141+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"profile":"core","run_id":"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"nemotron-nano-4b","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","parameters_b":null,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b","process_cpu_percent_avg":5.490567532375733,"process_rss_bytes_avg":5130473995.082253,"process_rss_bytes_peak":10141679616.0,"sample_count":1444,"sample_interval_ms":296.66646442619066,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6067978240.0}],"run":{"completed_at_utc":"2026-10-05T12:10:00.371141+00:00","created_at_utc":"2026-10-05T12:03:27.359922+00:00","git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","integrity_state":"VALID","manifest_checksum":"e0544884f17baca2f178756fd3cc437c5a49bf1769882d8b13c2d75abc1c2e7e","profile":"core","run_group":"2026-10-core-v1-nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0450422f-b932-4d66-a4df-836396550230","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"oos-calibration\"], \"model_keys\": [\"nemotron-nano-4b\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T12:03:27.359922+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d30bbd79-7c7e-4c87-b534-a4646ba86e89","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T12:03:27.373585+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":1074.4895419920795,"error_message":null,"error_type":null,"event_id":"2d05e24c-03bb-4cb8-880d-9abbd5a0df91","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T12:03:28.448660+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"294e8404-d75a-4330-a74e-1bc4297552f8","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T12:03:28.448880+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"069f1a8d-e408-4c69-a615-a8a65713ae2e","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T12:03:28.470038+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6ec45d09-265a-4983-ad7d-3414eb19ce63","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T12:09:59.965070+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1b6de5d8-a292-4af1-80a0-a8a612ecdfc5","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T12:09:59.965256+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":396.4203749928856,"error_message":null,"error_type":null,"event_id":"b119e14d-d789-49b0-a956-7c0547745710","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T12:10:00.361826+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3f0ce4b5-cb93-4e83-aed9-e61cba6cef3f","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T12:10:00.370975+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b306ca4d-ab91-44d2-ac02-3c0539dc53ae","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T12:10:00.371141+00:00"}]},"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core":{"cells":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T13:24:18.004156+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.551948051948052,"profile":"core","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"nemotron-nano-4b-q8","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","parameters_b":null,"provider_key":"korgis","quantization":"Q8_0","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b-q8","process_cpu_percent_avg":4.532544993975123,"process_rss_bytes_avg":5542900553.420978,"process_rss_bytes_peak":9062940672.0,"sample_count":1798,"sample_interval_ms":315.6007340511925,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5441241088.0}],"run":{"completed_at_utc":"2026-10-05T13:24:18.004156+00:00","created_at_utc":"2026-10-05T13:15:27.324284+00:00","git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","integrity_state":"VALID","manifest_checksum":"e97faf653c57f2ffd26f45e9a484f42f325b7b972cd9f406ab1a74a0e94239eb","profile":"core","run_group":"2026-10-core-v1-nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2dfa0a51-04b7-40d9-9cbc-d7fce2e60e12","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"intent-classification\"], \"model_keys\": [\"nemotron-nano-4b-q8\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T13:15:27.324284+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"75156827-1a4b-4453-a46a-215be9d9e42a","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:15:27.341964+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":1061.6437500139,"error_message":null,"error_type":null,"event_id":"11e343a3-d43b-4daa-8f33-e22f97e2c9d9","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:15:28.402943+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b26a2e7e-8866-47bd-a499-39259bb7f477","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:15:28.403221+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ca869566-b0ba-46fc-a404-70a086f10075","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:15:28.409754+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8721b9a8-c0f1-4b02-bd39-9ab30e1a62bf","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:24:17.480285+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7dd3003c-3bec-40ba-a273-e919055244b9","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:24:17.480985+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":503.02791700232774,"error_message":null,"error_type":null,"event_id":"89016dd7-f91d-47b8-8b1d-420376c718d6","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:24:17.984190+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"25840c1b-bd58-490e-9ace-48f070d846e3","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:24:18.004010+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6049a65b-93c2-4ff2-ab67-7900c6f48124","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T13:24:18.004156+00:00"}]},"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core":{"cells":[{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T13:15:22.380707+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.6,"profile":"core","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"nemotron-nano-4b-q8","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","parameters_b":null,"provider_key":"korgis","quantization":"Q8_0","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b-q8","process_cpu_percent_avg":8.601215146103618,"process_rss_bytes_avg":7892413781.333334,"process_rss_bytes_peak":11438456832.0,"sample_count":158,"sample_interval_ms":252.1124605016667,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":7386234880.0}],"run":{"completed_at_utc":"2026-10-05T13:15:22.380707+00:00","created_at_utc":"2026-10-05T13:14:49.110793+00:00","git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","integrity_state":"VALID","manifest_checksum":"3dca2d5a6acfab1ae30dc8b72b6507496fbde3cdbedfe8bc1f7f68d05e814541","profile":"core","run_group":"2026-10-core-v1-nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"924e3a7b-dee2-420c-b91e-2a688a5e9769","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"model_keys\": [\"nemotron-nano-4b-q8\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T13:14:49.110793+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"60c6b41b-d887-42bc-8fd4-e62e5f6e683f","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:14:49.123600+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":1070.9253339882707,"error_message":null,"error_type":null,"event_id":"efe662ee-02ef-4d33-933c-c11313fcd875","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:14:50.194963+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ebfdb1e4-235d-4b05-a2b4-fbfc058f571f","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:14:50.195275+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2c275298-9595-40bd-a6be-b75a5dd14eb3","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:14:50.205734+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e542395b-d483-45f5-a259-a10e859a0024","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:15:22.038768+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a4ea91e4-a678-4243-b2dd-23ff5df32963","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:15:22.038947+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":336.6249590035295,"error_message":null,"error_type":null,"event_id":"aa79377b-7a34-4fc0-b84d-7d58559f174b","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:15:22.375713+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"9f369370-af27-4470-b570-1de12e8488ab","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:15:22.380482+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"41c2a06c-5732-4730-ad69-33c89eb1e4bd","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 40, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 40, \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T13:15:22.380707+00:00"}]},"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core":{"cells":[{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T13:32:50.472664+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"profile":"core","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"nemotron-nano-4b-q8","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","parameters_b":null,"provider_key":"korgis","quantization":"Q8_0","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b-q8","process_cpu_percent_avg":5.151403915249351,"process_rss_bytes_avg":5654897500.709869,"process_rss_bytes_peak":8370143232.0,"sample_count":1747,"sample_interval_ms":306.91440564565875,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5491818496.0}],"run":{"completed_at_utc":"2026-10-05T13:32:50.472664+00:00","created_at_utc":"2026-10-05T13:24:26.151425+00:00","git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","integrity_state":"VALID","manifest_checksum":"1a4c62a72b9255581cebe7de712cbe473d0901131ab577415c23ca2bbb5f9a65","profile":"core","run_group":"2026-10-core-v1-nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ab5aa304-399b-4930-a058-36bae05a7730","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"oos-calibration\"], \"model_keys\": [\"nemotron-nano-4b-q8\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T13:24:26.151425+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"aae37536-9dcc-4e7c-85fb-307ac287617f","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:24:26.171142+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2103.908541990677,"error_message":null,"error_type":null,"event_id":"f30fb65a-ad30-474d-9b31-dc0b7c7c0599","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:24:28.275588+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"218a4dca-bc68-48a7-8627-8df9302a6674","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:24:28.275849+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"de2c83fe-26b3-40f7-8472-4c039d18bbee","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:24:28.299576+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ddc3b08a-24d2-4a73-bb07-fb8ec2c70033","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:32:49.986117+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1aedb7db-fe3e-4cf6-bd70-aaadf1f53fa3","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:32:49.986670+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":463.82266699220054,"error_message":null,"error_type":null,"event_id":"046b2cbc-09c2-4f6e-956a-7fa91081c64a","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:32:50.450778+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ffeb2225-221b-4dc7-86bd-f56bc1730ae5","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:32:50.472506+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"041c3d25-3d5d-4888-a7dd-857719b8ba61","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T13:32:50.472664+00:00"}]},"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core":{"cells":[{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T13:14:43.428857+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"profile":"core","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"nemotron-nano-4b-q8","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","parameters_b":null,"provider_key":"korgis","quantization":"Q8_0","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b-q8","process_cpu_percent_avg":7.166483605707706,"process_rss_bytes_avg":6463349882.22984,"process_rss_bytes_peak":10168467456.0,"sample_count":359,"sample_interval_ms":297.72169937988014,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5578047488.0}],"run":{"completed_at_utc":"2026-10-05T13:14:43.428857+00:00","created_at_utc":"2026-10-05T13:13:07.763290+00:00","git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","integrity_state":"VALID","manifest_checksum":"aeb8000f9a1bef7483dcfd81b396f5fc971668ed834dcee0981a591c8ca21928","profile":"core","run_group":"2026-10-core-v1-nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"9b7f4bc9-9ce6-47c1-ace7-d011780ecf10","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"qa-abstention\"], \"model_keys\": [\"nemotron-nano-4b-q8\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T13:13:07.763290+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"729c303d-4b6d-4dea-9a70-1233aeb2a190","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:13:07.782726+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2102.4610840104287,"error_message":null,"error_type":null,"event_id":"6d8e6c83-1a01-4178-bd4d-5008a6c6f771","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:13:09.885858+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1452d2fc-3381-44d4-ba87-91f723c14ea5","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:13:09.886130+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b5b1e52a-7612-4829-b6f9-3af0caa8e259","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:13:09.909526+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0a6e6f21-55a6-461c-a655-bd051bd52d36","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:14:42.981200+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8d7c145e-b9ad-442c-ac5f-a7826c555424","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:14:42.981387+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":441.9684159947792,"error_message":null,"error_type":null,"event_id":"3624f4d7-e188-4ce4-89c2-ddd85cf680dc","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:14:43.423519+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ed5e080d-702b-4175-a047-725767e26ed6","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:14:43.428667+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6f0ac3ce-564a-4ab0-91b5-ec73096a5ad4","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T13:14:43.428857+00:00"}]},"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core":{"cells":[{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T13:13:01.309902+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7166666666666667,"profile":"core","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"nemotron-nano-4b-q8","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","parameters_b":null,"provider_key":"korgis","quantization":"Q8_0","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b-q8","process_cpu_percent_avg":4.8684869265121,"process_rss_bytes_avg":5683850551.952017,"process_rss_bytes_peak":6735822848.0,"sample_count":654,"sample_interval_ms":307.429391476698,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5616762880.0}],"run":{"completed_at_utc":"2026-10-05T13:13:01.309902+00:00","created_at_utc":"2026-10-05T13:09:50.604381+00:00","git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","integrity_state":"VALID","manifest_checksum":"1e0de1e13f7df67283f054cc558bde51a2ef0b8c6e02642c871a333eedda714b","profile":"core","run_group":"2026-10-core-v1-nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"96924d71-07d8-4507-885c-6182c3221fd1","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"structured-output\"], \"model_keys\": [\"nemotron-nano-4b-q8\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T13:09:50.604381+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1ed15b37-2167-4a4d-9768-c033edfe79f4","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:09:50.634820+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":3107.3147090064595,"error_message":null,"error_type":null,"event_id":"3fdfac36-7a4b-412f-956e-9be7bf56e904","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:09:53.744040+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"4616b907-b789-47b3-8fe1-13d7c5b7e491","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:09:53.744363+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e6add32b-7137-43fe-a2e7-30d38e765e3d","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:09:53.795478+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a4db84bb-944e-446e-bbcc-f4df1ebc507b","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:13:00.691825+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"18c825a6-17e9-4830-b7d2-4fabb8d3760e","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:13:00.692084+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":608.8040419999743,"error_message":null,"error_type":null,"event_id":"fd1339dd-2b9b-4c37-a4b3-6f224e4b275e","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:13:01.301054+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ade389d7-178b-4f8b-a653-75cc3fe6e94c","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core\"}","model_key":"nemotron-nano-4b-q8","run_id":"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:13:01.309718+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d17d9c1d-135c-412f-bf7e-3c1501fc290f","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"2026-10-core-v1-nemotron-nano-4b-q8\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-q8-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T13:13:01.309902+00:00"}]},"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core":{"cells":[{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T11:57:09.469183+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"profile":"core","run_id":"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"nemotron-nano-4b","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","parameters_b":null,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b","process_cpu_percent_avg":9.533096212147692,"process_rss_bytes_avg":7103316869.12,"process_rss_bytes_peak":11089903616.0,"sample_count":264,"sample_interval_ms":273.58964924757555,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6888439808.0}],"run":{"completed_at_utc":"2026-10-05T11:57:09.469183+00:00","created_at_utc":"2026-10-05T11:56:08.892090+00:00","git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","integrity_state":"VALID","manifest_checksum":"bb0978ab456999693cb523f73a8887d279ea593fc6c2880a22b4a39a853b98bc","profile":"core","run_group":"2026-10-core-v1-nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f47bcb21-0ab9-4ef9-a0e7-a3851ed8f42c","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"qa-abstention\"], \"model_keys\": [\"nemotron-nano-4b\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T11:56:08.892090+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f4ca6050-8460-4c85-b6e9-ffb0f9e8453a","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:56:08.912737+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":1096.5186660032487,"error_message":null,"error_type":null,"event_id":"6583fad9-75a9-4bfd-8a49-97f540eeecc4","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:56:10.009605+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"64a1bda8-c1cc-47a1-8be3-7058969646e0","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:56:10.009947+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"9562cb68-6b96-4cca-ac5b-78b35992e381","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:56:10.032587+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"569f87c1-7e8a-4f7e-ab90-6d9d4c28e683","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:57:09.180433+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"cd8f82a9-b4df-4608-9a2b-82cb5c8e7e71","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:57:09.180616+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":281.7744589992799,"error_message":null,"error_type":null,"event_id":"a557759b-491a-4942-8fa6-a5caf1b84ce9","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:57:09.462527+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"fd7fe670-fb90-4b1b-9e76-6a06ef71c5c2","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:57:09.469041+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"cd6294d9-9b5c-4fd8-b5ce-963a8c6874a0","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T11:57:09.469183+00:00"}]},"2026-10-core-v1-nemotron-nano-4b-structured-output-core":{"cells":[{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T11:56:03.800158+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7333333333333333,"profile":"core","run_id":"2026-10-core-v1-nemotron-nano-4b-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-structured-output-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"nemotron-nano-4b","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","parameters_b":null,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"2026-10-core-v1-nemotron-nano-4b-structured-output-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"nemotron-nano-4b","process_cpu_percent_avg":8.190733436222855,"process_rss_bytes_avg":4822041698.065609,"process_rss_bytes_peak":6531989504.0,"sample_count":423,"sample_interval_ms":283.9724207012575,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5921701888.0}],"run":{"completed_at_utc":"2026-10-05T11:56:03.800158+00:00","created_at_utc":"2026-10-05T11:54:12.388074+00:00","git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","integrity_state":"VALID","manifest_checksum":"b2641e6eaa9eaa21768eb9b505c87fcf4e731b427d2adee63c1dec29d96bda35","profile":"core","run_group":"2026-10-core-v1-nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-structured-output-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f6c67a7d-049d-49d2-928f-2c2b5067c6f3","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"structured-output\"], \"model_keys\": [\"nemotron-nano-4b\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-structured-output-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T11:54:12.388074+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6c839a13-ac84-4316-94f7-e0d04a036304","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:54:12.412042+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":3110.5037499946775,"error_message":null,"error_type":null,"event_id":"3c00561e-1a11-49e6-a93d-4558a1124306","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:54:15.523002+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2e3485d7-9d3b-4095-a0f1-c0c7367ffc18","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:54:15.523374+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f9f1e6b3-69fd-4e70-885a-7a1dfb2fd156","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:54:15.574667+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c7c8d91f-f002-40e0-95e1-bc3a8da6b984","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:56:03.348435+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8c266457-9700-4703-ab10-f68ca76199de","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:56:03.348669+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":445.68562500353437,"error_message":null,"error_type":null,"event_id":"803c8791-0560-4e3c-acb8-86967a31a75f","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:56:03.794523+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"216778f6-aed6-461e-ae98-abbe8c26a62a","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-structured-output-core\"}","model_key":"nemotron-nano-4b","run_id":"2026-10-core-v1-nemotron-nano-4b-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:56:03.799979+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1d3ffc8e-9ee9-4325-8ee3-7a25d16da90d","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"2026-10-core-v1-nemotron-nano-4b\", \"run_id\": \"2026-10-core-v1-nemotron-nano-4b-structured-output-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-nemotron-nano-4b-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T11:56:03.800158+00:00"}]},"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core":{"cells":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T13:37:29.943647+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.2792207792207792,"profile":"core","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","parameters_b":null,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-0.8b-q4km","process_cpu_percent_avg":12.95914646037019,"process_rss_bytes_avg":4417102107.705627,"process_rss_bytes_peak":7946436608.0,"sample_count":462,"sample_interval_ms":199.48604018792173,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":10223632384.0}],"run":{"completed_at_utc":"2026-10-05T13:37:29.943647+00:00","created_at_utc":"2026-10-05T13:36:18.204979+00:00","git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","integrity_state":"VALID","manifest_checksum":"cc039ff3217383370504b4de6344993f91423246499e1d84314d8f973a5438dc","profile":"core","run_group":"2026-10-core-v1-qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c23aaa88-6594-4b91-8e19-10414c0fd7fe","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"intent-classification\"], \"model_keys\": [\"qwen3.5-0.8b-q4km\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T13:36:18.204979+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"98b19dae-d832-465b-9fe3-3c0b786b99f3","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:36:18.217662+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":1064.374375011539,"error_message":null,"error_type":null,"event_id":"0ee0060b-8c7c-40ce-9260-b8af84d042a6","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:36:19.282416+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"cfe3e66a-6c20-4e65-8bf7-2be629ab9021","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:36:19.282722+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"36346188-c671-4236-b07e-0c0fa40aa16d","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:36:19.289424+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3b9a1f9d-1877-485d-a532-7cceac504a37","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:37:29.638003+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a19fcf06-5c88-4f62-b875-f3175e0faf88","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:37:29.638288+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":294.05825000139885,"error_message":null,"error_type":null,"event_id":"fffa367a-14ed-4b66-ae63-91f0d12a5526","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:37:29.932523+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"77b7a5e1-e009-473b-94ab-7da4cb4a3ff7","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:37:29.943436+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8d2c79a8-a073-45b4-adbd-6e7e48d9ea97","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T13:37:29.943647+00:00"}]},"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core":{"cells":[{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T13:36:13.768283+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.2,"profile":"core","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","parameters_b":null,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-0.8b-q4km","process_cpu_percent_avg":18.636242311523596,"process_rss_bytes_avg":1745210777.6,"process_rss_bytes_peak":2670067712.0,"sample_count":81,"sample_interval_ms":213.9264905885284,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":15259779072.0}],"run":{"completed_at_utc":"2026-10-05T13:36:13.768283+00:00","created_at_utc":"2026-10-05T13:36:01.400213+00:00","git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","integrity_state":"VALID","manifest_checksum":"0ad33ce26fc18901b0a6e10892573771efd3f804b0f338e20e42108b591acd54","profile":"core","run_group":"2026-10-core-v1-qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"07bd399c-109c-4ffe-9580-5cff2579ea75","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"model_keys\": [\"qwen3.5-0.8b-q4km\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T13:36:01.400213+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0d65950e-5ec4-41b8-af6f-877210c5702a","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:36:01.421145+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":1076.1348329979228,"error_message":null,"error_type":null,"event_id":"95669def-8077-40da-a98b-5d182886f3b2","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:36:02.497705+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2e0000a8-e5a8-4e1d-b755-d1ba6dff86d8","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:36:02.498126+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"9cca0433-0914-4b5c-a5b2-4f3852477c01","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:36:02.510860+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"98ae42a9-ee7a-44a1-82ef-d86e8871d202","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:36:13.632959+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"58af0f09-55e9-4b0a-8b1f-0868fbfb9158","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:36:13.633170+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":131.1419580015354,"error_message":null,"error_type":null,"event_id":"f965a403-3c71-4bb9-a99b-763d13fb26af","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:36:13.764494+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6c9f722f-dac7-48be-bd2b-fe0129932ebd","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:36:13.768108+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"fccc2ada-6a90-4e3f-8cbc-c26002d6bf84","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 40, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 40, \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T13:36:13.768283+00:00"}]},"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core":{"cells":[{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T13:39:04.217888+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.5,"profile":"core","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","parameters_b":null,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-0.8b-q4km","process_cpu_percent_avg":16.11579774911694,"process_rss_bytes_avg":5182799526.2337675,"process_rss_bytes_peak":9496838144.0,"sample_count":463,"sample_interval_ms":247.8098196266941,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":9132261376.0}],"run":{"completed_at_utc":"2026-10-05T13:39:04.217888+00:00","created_at_utc":"2026-10-05T13:37:37.233096+00:00","git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","integrity_state":"VALID","manifest_checksum":"3bb15d59883261f0cbd8bc761a9408f96aa0b9bcb10c68d114595eb069045c7f","profile":"core","run_group":"2026-10-core-v1-qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e7913446-533e-4882-bd8d-c7d08b6ec0d7","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"oos-calibration\"], \"model_keys\": [\"qwen3.5-0.8b-q4km\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T13:37:37.233096+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c865b9a6-529b-42ed-a127-d20ce7042809","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:37:37.246216+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":1081.2740420078626,"error_message":null,"error_type":null,"event_id":"9f2f7de6-b4b7-442f-b80e-86525d1e8986","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:37:38.327878+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"95f19ed4-9dfe-4f5e-951a-b998d7011622","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:37:38.328237+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ebbdcb0e-a3f9-466b-aea6-992aaf80e275","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:37:38.349301+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"da956618-8ea1-4337-be3f-61b467c2141a","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:39:03.859701+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ffccaa14-ed15-4d10-8468-7844b7a8b35c","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:39:03.859860+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":348.0792499904055,"error_message":null,"error_type":null,"event_id":"28a66c88-4be0-44ae-8cc5-91950cd5ae15","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:39:04.208098+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3e2183fd-29b1-43a1-a6ad-c0e678ccfe82","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:39:04.217625+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b7efa6ab-68a0-421a-a9cc-3673ac4265d6","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T13:39:04.217888+00:00"}]},"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core":{"cells":[{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T13:35:56.623493+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.48333333333333334,"profile":"core","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","parameters_b":null,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-0.8b-q4km","process_cpu_percent_avg":16.023245275126758,"process_rss_bytes_avg":2421507094.7555556,"process_rss_bytes_peak":3926736896.0,"sample_count":126,"sample_interval_ms":259.9318482166079,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":14612348928.0}],"run":{"completed_at_utc":"2026-10-05T13:35:56.623493+00:00","created_at_utc":"2026-10-05T13:35:35.053360+00:00","git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","integrity_state":"VALID","manifest_checksum":"977d62d9698e56ac7f9d3b709f8952db7d6ff777bd8db092c233192c61645ac1","profile":"core","run_group":"2026-10-core-v1-qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"84fc4bee-3ea0-4c31-8216-13522d8f8b22","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"qa-abstention\"], \"model_keys\": [\"qwen3.5-0.8b-q4km\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T13:35:35.053360+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"049344e0-1ef0-4c9f-a18c-2b7773d274b9","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:35:35.075927+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":1082.855250002467,"error_message":null,"error_type":null,"event_id":"29ae3bd6-b470-4ca1-bd35-281fd97fffad","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:35:36.159201+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3de1f8e0-1755-459b-93f9-490b1a1e1d1c","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:35:36.159660+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3a33ae1c-029a-4fe9-8f7e-5a50f6200600","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:35:36.184845+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6338e2a7-778c-4c2d-bc96-52f1f0348834","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:35:56.489624+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ecd0e982-9d25-41f4-b008-c6b5a028ec11","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:35:56.489856+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":129.05258299724665,"error_message":null,"error_type":null,"event_id":"5f9a70aa-8034-42b5-aebe-b40aa3db541d","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:35:56.619083+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"abb6d6ff-d74c-4a9b-890a-a00647ecc4aa","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:35:56.623327+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7b7be593-cdb4-45bd-bfe5-132f629571ab","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T13:35:56.623493+00:00"}]},"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core":{"cells":[{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T13:35:30.102922+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.6833333333333333,"profile":"core","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","parameters_b":null,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-0.8b-q4km","process_cpu_percent_avg":20.24050747900639,"process_rss_bytes_avg":2703435980.8,"process_rss_bytes_peak":4575264768.0,"sample_count":195,"sample_interval_ms":237.40003127216673,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":13519536128.0}],"run":{"completed_at_utc":"2026-10-05T13:35:30.102922+00:00","created_at_utc":"2026-10-05T13:34:51.671219+00:00","git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","integrity_state":"VALID","manifest_checksum":"36e32c3ed3d237be04460dd4c019a8835f662097cee04d0f39596a993e279025","profile":"core","run_group":"2026-10-core-v1-qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e9debe14-6d2d-4e6a-b8b5-0d04bfa09bdf","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"structured-output\"], \"model_keys\": [\"qwen3.5-0.8b-q4km\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T13:34:51.671219+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7536f5ce-749a-41d3-b4ca-c15e579b1855","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:34:51.693396+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2102.7313340018736,"error_message":null,"error_type":null,"event_id":"67343cd3-81f2-4b67-a0ee-046b013368ff","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:34:53.796634+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d9e73ada-dfd4-481a-aa27-24263b78aa31","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:34:53.797161+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"562401cf-3ab0-47e2-8c59-cf2b0d7b63fd","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:34:53.867432+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"db4e4062-0727-4692-876b-b545fa8f600f","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:35:29.910280+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f821fdd4-a7f4-46a7-9ec2-04ebf721538f","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:35:29.910461+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":186.83574999158736,"error_message":null,"error_type":null,"event_id":"b93d7f16-19df-40e2-8c8c-328b7455533f","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:35:30.097553+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ff0ecc07-594e-403e-82b4-6762530dc6fe","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:35:30.102734+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c537b94d-4b27-4a34-acb3-51aadf6bece1","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"2026-10-core-v1-qwen3.5-0.8b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-0.8b-q4km-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T13:35:30.102922+00:00"}]},"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core":{"cells":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T11:25:37.855179+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.577922077922078,"profile":"core","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","parameters_b":null,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-2b-q4km","process_cpu_percent_avg":9.421328984702535,"process_rss_bytes_avg":2155211762.7012987,"process_rss_bytes_peak":3131637760.0,"sample_count":616,"sample_interval_ms":260.0283379352414,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6455427072.0}],"run":{"completed_at_utc":"2026-10-05T11:25:37.855179+00:00","created_at_utc":"2026-10-05T11:23:23.440620+00:00","git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","integrity_state":"VALID","manifest_checksum":"158de829915981d818d2d06f0126a88ef073c7471d86d3a7e37bcc2ebb3ad099","profile":"core","run_group":"2026-10-core-v1-qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a9a238c5-3dda-4986-9de2-d2bb8e1e2a3d","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"intent-classification\"], \"model_keys\": [\"qwen3.5-2b-q4km\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T11:23:23.440620+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e16bb27c-4b8f-4b7c-a276-a4eca576f86c","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:23:23.466516+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2108.304584005964,"error_message":null,"error_type":null,"event_id":"f42adc81-0cf4-485b-8ffc-c64d4e52db8e","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:23:25.575613+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6d2a710b-175b-4ff7-9756-02628cbae46a","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:23:25.576045+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e8a8cea2-204b-4ae5-a98c-404c0422048c","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:23:25.585745+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c35b554a-e046-4cac-85ff-615c9475db77","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:25:37.283542+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"46e25378-a82e-4ab4-8faa-cfd70e6c5f3d","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:25:37.283795+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":557.1215829986613,"error_message":null,"error_type":null,"event_id":"2e3ee5da-e8f6-481a-b7e3-3833fb7a001b","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:25:37.841120+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"5902a4d7-8226-4e9d-a9fe-81f8046ddf8f","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:25:37.854954+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7962f5d1-8bba-4d8a-95f8-95ffcd67cbfa","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T11:25:37.855179+00:00"}]},"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core":{"cells":[{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T11:23:17.557837+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.325,"profile":"core","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","parameters_b":null,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-2b-q4km","process_cpu_percent_avg":12.866597281256125,"process_rss_bytes_avg":2496888695.466666,"process_rss_bytes_peak":3418226688.0,"sample_count":104,"sample_interval_ms":243.8169525759804,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":8507686912.0}],"run":{"completed_at_utc":"2026-10-05T11:23:17.557837+00:00","created_at_utc":"2026-10-05T11:22:57.966728+00:00","git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","integrity_state":"VALID","manifest_checksum":"b3c4c4880aecfd061ade8a343253ebad8374c8334ab741017ae12d5cd6a944d6","profile":"core","run_group":"2026-10-core-v1-qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"21e2f61e-e022-44ab-9787-d1d890425c58","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"model_keys\": [\"qwen3.5-2b-q4km\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T11:22:57.966728+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8061c1ac-c87f-4324-bf79-cc8c7db74ec4","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:22:57.985524+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2106.1506669939263,"error_message":null,"error_type":null,"event_id":"d75ef71d-a5d8-4a81-aee5-30ee0b82c51a","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:23:00.092356+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"93786d76-794d-44ff-b884-1f638da24251","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:23:00.092738+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3999e84a-97e3-4322-8b01-502bf4d446f0","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:23:00.107330+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"579e03cb-b0e5-4446-b468-40837c240614","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:23:17.373125+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f8603038-0cc8-4e6a-9813-d8ca4587ca4f","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:23:17.373371+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":179.08533300214913,"error_message":null,"error_type":null,"event_id":"b6065024-4036-4c3a-b132-a041491b3674","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:23:17.552652+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f66952a0-0c54-4a57-bb73-4201c3412079","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:23:17.557643+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0a73da56-9d3c-4f70-8a1f-ecd05ade8395","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 40, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 40, \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T11:23:17.557837+00:00"}]},"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core":{"cells":[{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T11:28:16.098461+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.5,"profile":"core","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","parameters_b":null,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-2b-q4km","process_cpu_percent_avg":8.346543248013683,"process_rss_bytes_avg":3164761335.3558445,"process_rss_bytes_peak":6107152384.0,"sample_count":621,"sample_interval_ms":291.97748509630554,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6536364032.0}],"run":{"completed_at_utc":"2026-10-05T11:28:16.098461+00:00","created_at_utc":"2026-10-05T11:25:47.402412+00:00","git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","integrity_state":"VALID","manifest_checksum":"fd5afbc9cec0789853b013165eeb023ea7df66f411e9f6a0ab94babf0e1fc9ea","profile":"core","run_group":"2026-10-core-v1-qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7e6e3661-24c0-40da-b88f-05cebba69eea","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"oos-calibration\"], \"model_keys\": [\"qwen3.5-2b-q4km\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T11:25:47.402412+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"fdf000a6-c870-4d8c-bc42-233799bbffac","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:25:47.426722+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2109.4925420038635,"error_message":null,"error_type":null,"event_id":"8d7be049-7283-4ab2-ba4a-50ef71fdfb7b","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:25:49.536865+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"bea0f489-c291-499c-a994-e3ebf15f233f","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:25:49.537360+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0f812f6d-db02-420f-9a43-5c49b35a7f10","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:25:49.568994+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"158a7380-a26a-40cd-8342-bebb87fa7e62","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:28:15.688710+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"641adba7-27f4-420f-933c-a6723a42688e","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:28:15.688992+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":400.3344590018969,"error_message":null,"error_type":null,"event_id":"3efca248-5c79-456d-b2b9-3524ea553ef2","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:28:16.089631+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"4d2cc032-058f-483b-8afc-3f73b8fb328f","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:28:16.098262+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"39ef9f8d-609f-4f49-939a-f43863d20d5f","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T11:28:16.098461+00:00"}]},"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core":{"cells":[{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T11:22:51.387753+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.7833333333333333,"profile":"core","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","parameters_b":null,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-2b-q4km","process_cpu_percent_avg":14.926939453226597,"process_rss_bytes_avg":2006719510.7555552,"process_rss_bytes_peak":2334687232.0,"sample_count":182,"sample_interval_ms":249.7748705191043,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6527320064.0}],"run":{"completed_at_utc":"2026-10-05T11:22:51.387753+00:00","created_at_utc":"2026-10-05T11:22:12.902724+00:00","git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","integrity_state":"VALID","manifest_checksum":"476644142ad30f9e096bff152179478a42842412e5cb2447f1e21087da186714","profile":"core","run_group":"2026-10-core-v1-qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"57213d9e-16d0-441b-bfb7-40e39e972dd1","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"qa-abstention\"], \"model_keys\": [\"qwen3.5-2b-q4km\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T11:22:12.902724+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b2799e55-a55c-4760-81ac-f0ae7bd19ccf","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:22:12.948944+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":3149.2560830083676,"error_message":null,"error_type":null,"event_id":"7a7147f5-9dd3-49c9-904a-7023042fd506","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:22:16.101303+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"37be3e13-1f71-4146-b79e-d699fbec0a2d","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:22:16.101681+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0864d9d6-7d57-491c-95bf-0a37b873eef9","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:22:16.130859+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"632519dc-f7b5-4f84-acbb-3dc9a48caded","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:22:51.091572+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c894bd5e-56d0-4ac9-8e61-a8fa97ea5033","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T11:22:51.091814+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":286.9867919944227,"error_message":null,"error_type":null,"event_id":"25a1b957-b2a6-4cb3-b6ae-1e17b8cb0119","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:22:51.379003+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8640cc33-7d4c-4b21-a669-a2ab5fef0d41","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T11:22:51.387482+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d3a6d275-a9d9-48ad-a35c-e72736d297ad","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T11:22:51.387753+00:00"}]},"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core":{"cells":[{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T10:55:49.921242+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7333333333333333,"profile":"core","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","parameters_b":null,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-2b-q4km","process_cpu_percent_avg":15.462938913292911,"process_rss_bytes_avg":2241664864.711111,"process_rss_bytes_peak":2811346944.0,"sample_count":253,"sample_interval_ms":275.0343119529134,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6283788288.0}],"run":{"completed_at_utc":"2026-10-05T10:55:49.921242+00:00","created_at_utc":"2026-10-05T10:54:47.306893+00:00","git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","integrity_state":"VALID","manifest_checksum":"315804e522753d719add5ff233955b59b1e96fd4bde073c96789f5a5d4dfce33","profile":"core","run_group":"2026-10-core-v1-qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"013ed6ee-f382-45f3-afb4-9b6fee0f4fa6","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"structured-output\"], \"model_keys\": [\"qwen3.5-2b-q4km\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T10:54:47.306893+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"651f0a39-1b96-43fb-83b6-e49c45480416","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T10:54:47.336516+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":3113.9263340010075,"error_message":null,"error_type":null,"event_id":"902dd87c-db7a-43f1-bbac-cdc3ec12f547","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T10:54:50.451018+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a043cb52-5987-4f08-941c-0176d8a939f5","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T10:54:50.517938+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ba57a648-f8a8-4694-9770-35a4317814e5","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T10:55:49.563192+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"68a6a0a6-397e-453f-9b0c-09d7f4ade3cc","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T10:55:49.563403+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":348.8127919990802,"error_message":null,"error_type":null,"event_id":"7808eed8-d216-4c19-a319-18d97704abe0","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T10:55:49.912426+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"162a2094-48fc-4dcf-82a5-8d047b6c17f9","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core\"}","model_key":"qwen3.5-2b-q4km","run_id":"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T10:55:49.921013+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"5d7649fd-a86b-43b0-bdbc-0c4ef9a71180","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"2026-10-core-v1-qwen3.5-2b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-2b-q4km-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T10:55:49.921242+00:00"}]},"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core":{"cells":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T12:46:44.225239+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6948051948051948,"profile":"core","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","parameters_b":null,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-9b-q4km","process_cpu_percent_avg":3.9138417761512048,"process_rss_bytes_avg":6965862817.821014,"process_rss_bytes_peak":10559422464.0,"sample_count":1620,"sample_interval_ms":303.4210088575144,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5489934336.0}],"run":{"completed_at_utc":"2026-10-05T12:46:44.225239+00:00","created_at_utc":"2026-10-05T12:39:07.425464+00:00","git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","integrity_state":"VALID","manifest_checksum":"f3b69d33b7282b4f52ea917cc13281b466d3aeeacba0850200180389a1efedab","profile":"core","run_group":"2026-10-core-v1-qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f12cd1db-6108-4c1d-991b-ce3d378881c8","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"intent-classification\"], \"model_keys\": [\"qwen3.5-9b-q4km\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T12:39:07.425464+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"993a6ce9-6c1b-4a55-af7d-a152645d0839","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T12:39:07.447726+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2088.1293330021435,"error_message":null,"error_type":null,"event_id":"f8bc9bbb-068d-4921-9727-06a1cf14ae65","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T12:39:09.536349+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d1fe9540-6e00-4b55-a00a-4dfd6020d809","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T12:39:09.536678+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7e3d8696-42e9-4951-a093-0cf10b325114","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T12:39:09.544191+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a55a1479-c416-478c-8c8b-990a079a097c","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T12:46:43.588498+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b74ab668-acc2-46eb-b031-742d7c76bcf7","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T12:46:43.589112+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":616.3298329920508,"error_message":null,"error_type":null,"event_id":"6260cf72-f358-4a2e-a1d0-5fbb346ad6bc","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T12:46:44.205633+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b9a69502-936f-4b1f-bae8-c4eb912523b5","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T12:46:44.225050+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ef313388-0e80-4133-91d2-8e31fea7e1ae","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T12:46:44.225239+00:00"}]},"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core":{"cells":[{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T12:39:03.216607+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.675,"profile":"core","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","parameters_b":null,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-9b-q4km","process_cpu_percent_avg":6.526783901766963,"process_rss_bytes_avg":8611465482.24,"process_rss_bytes_peak":11088855040.0,"sample_count":175,"sample_interval_ms":275.92921112542774,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":11636277248.0}],"run":{"completed_at_utc":"2026-10-05T12:39:03.216607+00:00","created_at_utc":"2026-10-05T12:38:21.775724+00:00","git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","integrity_state":"VALID","manifest_checksum":"724ab7fc63e7eb9e8a7c3a0b1a4faeb461377b247fe7a608d6149af7525e2e33","profile":"core","run_group":"2026-10-core-v1-qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"331a7360-28e6-431b-94a6-f6c37968d49e","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"model_keys\": [\"qwen3.5-9b-q4km\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T12:38:21.775724+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2695e1df-abf0-4a1a-b642-45628d5f1c3a","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T12:38:21.797785+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2096.915208006976,"error_message":null,"error_type":null,"event_id":"9e99febf-f047-42a5-bc6e-fa665521d3fc","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T12:38:23.895135+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d47b82d0-4258-4a34-aa0c-4dc36380fa14","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T12:38:23.895433+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"aa643d69-0227-4f5e-bc82-a3b5bd9a0b82","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T12:38:23.907199+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3079adc2-0638-4237-a730-e37e95086f51","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T12:39:03.084104+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"aa7da1af-23f5-4497-87b9-485910dbc6a6","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T12:39:03.084264+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":129.61845900281332,"error_message":null,"error_type":null,"event_id":"efd8173b-553d-4c1b-b5cc-bf21dc2a9ada","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T12:39:03.213999+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e8bb0fac-8575-46e6-b117-2f8d4953dbe7","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T12:39:03.216489+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"be5c8629-eb19-4067-aced-94bc4f483267","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 40, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 40, \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T12:39:03.216607+00:00"}]},"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core":{"cells":[{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T13:01:17.635848+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.974025974025974,"profile":"core","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","parameters_b":null,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-9b-q4km","process_cpu_percent_avg":2.683081412939199,"process_rss_bytes_avg":6844874999.319381,"process_rss_bytes_peak":10287939584.0,"sample_count":2848,"sample_interval_ms":314.6654996565356,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5142609920.0}],"run":{"completed_at_utc":"2026-10-05T13:01:17.635848+00:00","created_at_utc":"2026-10-05T12:46:51.362308+00:00","git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","integrity_state":"VALID","manifest_checksum":"40b11821dba58df8cbea4bf432dbe81d367768a1c069219ef9b3fd06679e04c7","profile":"core","run_group":"2026-10-core-v1-qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"455be3bc-14b2-4d9a-a0cd-07682acbbd0c","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"oos-calibration\"], \"model_keys\": [\"qwen3.5-9b-q4km\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T12:46:51.362308+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"77f702b8-5520-4b56-a35d-c60bf5688377","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T12:46:51.377464+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2098.7562499940395,"error_message":null,"error_type":null,"event_id":"2a5d31a4-19b9-456c-8005-32bdb198ef99","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T12:46:53.476781+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b27801a9-3424-44ae-9e7d-2fc405af45ce","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T12:46:53.477228+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"85401107-955e-4fd4-bcf7-57fe92777ba0","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T12:46:53.504418+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ff4e8242-6152-4d2d-a2f5-28f4373906c6","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:01:17.102425+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"4812b051-0144-485b-9d27-b809e5d5e0d9","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T13:01:17.103633+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":506.34112500119954,"error_message":null,"error_type":null,"event_id":"c48ac699-feec-43fe-96d9-190944658ac9","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:01:17.610177+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"eca53f61-327c-429c-8838-f2417a45d1c4","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T13:01:17.635658+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1dc5e595-f43a-4650-8c10-a1384e867b1b","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T13:01:17.635848+00:00"}]},"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core":{"cells":[{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T12:38:16.531029+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.9333333333333333,"profile":"core","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","parameters_b":null,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-9b-q4km","process_cpu_percent_avg":6.3201481230234045,"process_rss_bytes_avg":8711409521.940317,"process_rss_bytes_peak":11538563072.0,"sample_count":345,"sample_interval_ms":281.76636477524414,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":7018201088.0}],"run":{"completed_at_utc":"2026-10-05T12:38:16.531029+00:00","created_at_utc":"2026-10-05T12:36:50.317204+00:00","git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","integrity_state":"VALID","manifest_checksum":"b31ef91e87818e3a52b034d70c5f2b55c62f144e4bb9ae1f8cb95cdb654479fd","profile":"core","run_group":"2026-10-core-v1-qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3a8bf336-06ab-4d20-94cc-971e578a00b6","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"qa-abstention\"], \"model_keys\": [\"qwen3.5-9b-q4km\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T12:36:50.317204+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"982681d4-11a6-4e3e-b0ee-782d3d6ebd3f","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T12:36:50.339381+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2093.5112080042018,"error_message":null,"error_type":null,"event_id":"a42af8e9-3b23-4014-9ac1-be49d8eeb521","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T12:36:52.433363+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"aded9a82-436c-4d8e-bcd7-dc71ed19e1db","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T12:36:52.433675+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e9065788-18eb-4669-9fa2-6f2e2ea33008","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T12:36:52.458387+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"49b75f2f-9f3f-4523-92a8-186aa45feac1","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T12:38:16.128544+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"96624d19-9259-4a41-9d37-fcfdad355811","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T12:38:16.128784+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":396.42016601283103,"error_message":null,"error_type":null,"event_id":"66830d04-d1ca-40be-8e6b-549f90b411d8","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T12:38:16.525412+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"bc02f48d-b208-490d-90f0-bb07573f914a","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T12:38:16.530883+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"37858f2f-5633-4a23-a69a-a2d8dec4593b","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T12:38:16.531029+00:00"}]},"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core":{"cells":[{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T12:36:44.930375+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7666666666666667,"profile":"core","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","parameters_b":null,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-9b-q4km","process_cpu_percent_avg":5.8239325570441824,"process_rss_bytes_avg":6475378858.969259,"process_rss_bytes_peak":7243612160.0,"sample_count":617,"sample_interval_ms":298.72389312365516,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5452595200.0}],"run":{"completed_at_utc":"2026-10-05T12:36:44.930375+00:00","created_at_utc":"2026-10-05T12:33:48.470044+00:00","git_commit":"c934ac60383688117dfe2b0da8aff5941e6df0d1","integrity_state":"VALID","manifest_checksum":"f39f3f8f27b1efce332d14eb590a87c49c42d9985242af01bbf467bbf86dd556","profile":"core","run_group":"2026-10-core-v1-qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"38aa46fc-c9e7-443a-b1fc-38da83bb1ddb","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"structured-output\"], \"model_keys\": [\"qwen3.5-9b-q4km\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T12:33:48.470044+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"193626c0-37fa-4acb-bd3d-d01411cc1673","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T12:33:48.492599+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":5244.7962090082,"error_message":null,"error_type":null,"event_id":"05c38615-442f-42c2-9301-fa563af5c67e","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T12:33:53.737957+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7a332d48-2205-49d6-878a-1bbc6a8a44ac","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T12:33:53.738527+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"08c32fad-2757-44eb-b45e-eceff21dfa07","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T12:33:53.807990+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e20af737-03e6-4a61-bd28-2215a096d6ce","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T12:36:44.421238+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"fc18bc0f-91b1-45bb-ae48-c2b3b24db6e0","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T12:36:44.421451+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":502.32916700770147,"error_message":null,"error_type":null,"event_id":"758b042f-1a26-4d48-913a-b68b26167e37","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T12:36:44.923957+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"000957ff-7a34-40ed-b6d0-f232c5e2f110","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core\"}","model_key":"qwen3.5-9b-q4km","run_id":"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T12:36:44.930183+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e4c69b9c-c066-4957-827d-be3787ad6604","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"2026-10-core-v1-qwen3.5-9b-q4km\", \"run_id\": \"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-qwen3.5-9b-q4km-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T12:36:44.930375+00:00"}]},"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core":{"cells":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-06T14:37:14.912068+00:00","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","failure_count":0,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.5844155844155844,"profile":"core","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":6192222208,"deployment":"local","effective_model_id":"qwen3.8-27b-iq1s","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","family":"qwen3.8","model_id":"Qwen/Qwen3.8-27B","model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","parameters_b":27.0,"provider_key":"korgis","quantization":"UD-IQ1_S","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","model_key":"qwen3.8-27b-iq1s","process_cpu_percent_avg":3.5068517339810183,"process_rss_bytes_avg":8038410776.168479,"process_rss_bytes_peak":12158550016.0,"sample_count":4560,"sample_interval_ms":309.607195745226,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5166546944.0}],"run":{"completed_at_utc":"2026-10-06T14:37:14.912068+00:00","created_at_utc":"2026-10-06T13:56:23.026913+00:00","git_commit":"640138c3429882a42797d28074f98d3fea98a63b","integrity_state":"VALID","manifest_checksum":"8c4140cb9378a0910649f6682aedc5884cff3edcc55282dd089260f9f05b08a9","profile":"core","run_group":"2026-10-core-v1-qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"fce062ec-4bc9-4e15-a6e0-6096efe8b492","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"intent-classification\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"qwen3.8-27b-iq1s\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T13:56:23.026913+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c6efe32e-04a6-4a07-be04-e24ba2b5dc1e","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T13:56:23.049319+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2088.495125004556,"error_message":null,"error_type":null,"event_id":"1b6a451b-caa3-49b4-9691-d027a8373f62","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T13:56:25.138237+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3e2849a0-9619-405b-87e2-95fde007dae5","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T13:56:25.138522+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f09c4181-135d-40a7-98fc-f49c194275bf","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T13:56:25.144670+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d1c61517-caa3-4f6b-8522-e2842a7805bb","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T14:19:18.511077+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"fc46e0cf-76c0-46e4-b247-48211a2e8643","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T14:19:18.511686+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":568.7633749912493,"error_message":null,"error_type":null,"event_id":"d0cddaae-65f4-4fe8-b6a3-a95d71bc6cf7","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T14:19:19.080643+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1bedaaa8-a003-4bd6-b5d9-066572c34dfd","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T14:19:19.102709+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"41792f81-f2df-4e9c-bcee-30c0ba29b019","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T14:19:19.102854+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"97ea3dd5-7602-4e75-803b-61836e3804a8","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"intent-classification\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"qwen3.8-27b-iq1s\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T14:37:12.634279+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"092bc2f0-87f8-46b5-9918-eab3432d5d3c","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T14:37:12.654656+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2086.0242920025485,"error_message":null,"error_type":null,"event_id":"3e33b10d-c1f9-432f-865b-12dd8e280ca1","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T14:37:14.741133+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"229d4e35-15b5-4256-967f-515a5c68651a","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T14:37:14.741477+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"db2c22dd-7b50-4278-ad0e-aa7f06addfa3","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T14:37:14.748653+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0d95e5ae-b21e-4a8a-b1f6-42e94275d46a","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T14:37:14.773177+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"838a7c5f-9493-4561-82da-787d9a375685","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T14:37:14.773350+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":125.61029200151097,"error_message":null,"error_type":null,"event_id":"da1f990d-fcbf-49b8-8867-2cfe98102a93","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T14:37:14.899159+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"dadeeb5e-3303-4f64-98a1-5404b95e0edb","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T14:37:14.911885+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"91ce56b9-60e6-4936-81f7-9cbd5d9b3d65","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 0, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core\", \"skipped_cases\": 154}","model_key":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T14:37:14.912068+00:00"}]},"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core":{"cells":[{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-06T14:37:12.067024+00:00","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","failure_count":0,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.25,"profile":"core","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":6192222208,"deployment":"local","effective_model_id":"qwen3.8-27b-iq1s","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","family":"qwen3.8","model_id":"Qwen/Qwen3.8-27B","model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","parameters_b":27.0,"provider_key":"korgis","quantization":"UD-IQ1_S","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","model_key":"qwen3.8-27b-iq1s","process_cpu_percent_avg":7.113533739353301,"process_rss_bytes_avg":9138898947.755379,"process_rss_bytes_peak":12174065664.0,"sample_count":717,"sample_interval_ms":297.72640370809415,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5485953024.0}],"run":{"completed_at_utc":"2026-10-06T14:37:12.067024+00:00","created_at_utc":"2026-10-06T13:52:52.485489+00:00","git_commit":"640138c3429882a42797d28074f98d3fea98a63b","integrity_state":"VALID","manifest_checksum":"183492bb4bb2543169a05be29371855b15224b826e839352ba5fdb9a916fa0d6","profile":"core","run_group":"2026-10-core-v1-qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"9c714f19-4bc9-43c8-baa1-9b95dfd880c3","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"qwen3.8-27b-iq1s\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T13:52:52.485489+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7cef37e2-2b74-4f1c-9943-a9d823e4fac5","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T13:52:52.504272+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2090.875500012771,"error_message":null,"error_type":null,"event_id":"5ab6c507-d03b-4e67-9ed1-c1d5ee666919","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T13:52:54.595537+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"35d1e0c4-fe77-4d7e-ad5b-06fba12a424a","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T13:52:54.595893+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2dd6e875-e8ed-46e7-8f53-995441579669","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T13:52:54.608011+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"95b6257d-1138-4c1d-b13b-451bd3e561fd","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T13:56:21.961022+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1da1e627-4b2e-4e03-92db-cde30eb0e564","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T13:56:21.961267+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":398.5383750114124,"error_message":null,"error_type":null,"event_id":"fbbcfe22-cde1-4f4c-b6ac-52c2973b1ffb","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T13:56:22.359983+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3729754c-abde-4fed-9144-4489d8230b3b","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T13:56:22.364088+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1c795e12-2a2d-4c6b-8204-c353fdb43cdd","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 40, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 40, \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T13:56:22.364232+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"5b56af7a-91c0-408c-b267-6f67bdd4042a","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"qwen3.8-27b-iq1s\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T14:37:09.793777+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"38cf396f-b146-40cf-a531-d166e5bc996e","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T14:37:09.816586+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2088.4326670056907,"error_message":null,"error_type":null,"event_id":"f6fb2952-04dc-4514-9588-7f0944398041","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T14:37:11.905432+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3b564bfb-9d3b-4f3e-9431-489a9dae5623","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T14:37:11.905807+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e90dce42-3caf-49e6-aae5-f0f85ee0b7da","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T14:37:11.917943+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"5a60819d-dec8-4312-80ec-c4e6f76ef152","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T14:37:11.928126+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2f7107ff-a889-438b-86fb-7f9acd76bb3f","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T14:37:11.928311+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":132.91149999713525,"error_message":null,"error_type":null,"event_id":"e98438ec-c919-4e17-9bb1-88025bf1a965","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T14:37:12.061389+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b348b98e-ae0d-4616-abed-16d7c946a7db","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T14:37:12.066863+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"64c9fafd-9d58-4379-b289-f66fa9f99044","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 0, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 40, \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core\", \"skipped_cases\": 40}","model_key":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T14:37:12.067024+00:00"}]},"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core":{"cells":[{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-06T14:53:01.287222+00:00","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","failure_count":0,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.8181818181818182,"profile":"core","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","runtime_json":"{}","sweep_id":null},"failure_summary":[{"count":1,"error_type":"TransportError","event_type":"model.release.failed","first_at":"2026-10-06T14:37:02.350119+00:00","last_at":"2026-10-06T14:37:02.350119+00:00"}],"models":[{"artifact_format":"gguf","artifact_size_bytes":6192222208,"deployment":"local","effective_model_id":"qwen3.8-27b-iq1s","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","family":"qwen3.8","model_id":"Qwen/Qwen3.8-27B","model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","parameters_b":27.0,"provider_key":"korgis","quantization":"UD-IQ1_S","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","model_key":"qwen3.8-27b-iq1s","process_cpu_percent_avg":3.3561674118938543,"process_rss_bytes_avg":8080732806.054676,"process_rss_bytes_peak":11618025472.0,"sample_count":6492,"sample_interval_ms":307.9710546548535,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5336121344.0}],"run":{"completed_at_utc":"2026-10-06T14:53:01.287222+00:00","created_at_utc":"2026-10-06T14:19:19.878401+00:00","git_commit":"640138c3429882a42797d28074f98d3fea98a63b","integrity_state":"VALID","manifest_checksum":"8950211d12e495ee42dcf5e9e0d6e341427e75d6e06c843ace26ec607f5d647d","profile":"core","run_group":"2026-10-core-v1-qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"4d0f7c31-5172-4006-84b2-50f5d965125b","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"oos-calibration\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"qwen3.8-27b-iq1s\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T14:19:19.878401+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"fe5f90a4-0633-4aec-9d41-cba19d6f24b9","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T14:19:19.896748+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2106.9387500028824,"error_message":null,"error_type":null,"event_id":"f78007e9-692a-496a-a957-7252456d0513","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T14:19:22.005396+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2f042558-17a5-4006-9bc7-ab96484952b4","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T14:19:22.005701+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"05d9f0dc-5c9b-4fd8-adc8-d0c34b0abc45","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T14:19:22.028666+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"89a45b98-6415-447d-91db-9fe1e71defa7","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T14:37:02.340977+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":"HTTP 409: {\"detail\":\"Model 'qwen3.8-27b-iq1s' has an active request.\"}","error_type":"TransportError","event_id":"b294e8ca-d5aa-4781-bb54-12cfc64a2e4c","event_type":"model.release.failed","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","sample_id":null,"status":"failed","timestamp_utc":"2026-10-06T14:37:02.350119+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ed3e6ff0-8c57-4e57-abbb-51ae7a56570c","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"oos-calibration\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"qwen3.8-27b-iq1s\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T14:37:15.599353+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b1f09190-0975-487e-8672-de5e69ff1a78","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T14:37:15.618927+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2087.446749996161,"error_message":null,"error_type":null,"event_id":"3c4f050f-540d-4e13-a63b-a66b204fa613","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T14:37:17.706776+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ad53bddf-70d2-4032-8a32-461567f83c43","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T14:37:17.707133+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"36b5b7ae-f489-4024-a56b-d794671d1967","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T14:37:17.729491+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"4bd6b2b8-ce38-420d-8bad-48753e5a8ec4","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T14:53:00.746889+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"fba6608e-d651-4d26-b80c-9cc02d387e55","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T14:53:00.747706+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":511.9724579999456,"error_message":null,"error_type":null,"event_id":"b4d87eff-7267-4e08-9c92-6220950864ad","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T14:53:01.259836+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e4d08d95-3c09-4daa-9efd-c1852f7610d8","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T14:53:01.286822+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"691b9ebf-f16c-42af-b978-b71598b58975","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 66, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core\", \"skipped_cases\": 88}","model_key":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T14:53:01.287222+00:00"}]},"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core":{"cells":[{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-06T14:37:09.228523+00:00","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","failure_count":0,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8,"profile":"core","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":6192222208,"deployment":"local","effective_model_id":"qwen3.8-27b-iq1s","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","family":"qwen3.8","model_id":"Qwen/Qwen3.8-27B","model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","parameters_b":27.0,"provider_key":"korgis","quantization":"UD-IQ1_S","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","model_key":"qwen3.8-27b-iq1s","process_cpu_percent_avg":5.983418081521953,"process_rss_bytes_avg":9550237266.862207,"process_rss_bytes_peak":13159825408.0,"sample_count":792,"sample_interval_ms":301.02958077118194,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5182652416.0}],"run":{"completed_at_utc":"2026-10-06T14:37:09.228523+00:00","created_at_utc":"2026-10-06T13:49:05.732763+00:00","git_commit":"640138c3429882a42797d28074f98d3fea98a63b","integrity_state":"VALID","manifest_checksum":"25b65d9047d5dfc634ad240d8cdf00e04ffb76829026f345e9f4a766b1bfbd02","profile":"core","run_group":"2026-10-core-v1-qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"193f66ff-6281-46b2-960a-20b799562ebe","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"qa-abstention\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"qwen3.8-27b-iq1s\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T13:49:05.732763+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c8fba094-5796-4052-b6a1-46052ff981e9","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T13:49:05.756224+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2102.0136670122156,"error_message":null,"error_type":null,"event_id":"a76b3ee2-d86f-4216-a3b5-11673a2f4cf0","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T13:49:07.858654+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"eed8a60a-0e1d-4611-8b6d-91caa4cb2769","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T13:49:07.858919+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"5937d989-712b-45a4-aa22-7f4caa172468","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T13:49:07.881910+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"00ccb9fc-f03a-4e58-b4ee-a89368a2e2f8","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T13:52:51.489329+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1fa95d5e-5063-4279-ac92-15f258079049","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T13:52:51.489614+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":393.6293750011828,"error_message":null,"error_type":null,"event_id":"41c47354-e748-4cbe-bb1e-a521e8dddeb5","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T13:52:51.883450+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"291266c3-d748-4144-898f-fc393662ee7a","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T13:52:51.889025+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"22079451-ed39-4679-b58c-f441373c47a0","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T13:52:51.889211+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c67ed667-2063-4e4d-a066-ca8b74691f3c","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"qa-abstention\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"qwen3.8-27b-iq1s\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T14:37:06.953279+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b26d10f9-0494-4e66-b674-b8c5c9610c71","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T14:37:06.975778+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2077.712500002235,"error_message":null,"error_type":null,"event_id":"70774b1e-5ead-4594-b263-c0151af6e825","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T14:37:09.053903+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8c8e68d9-9a10-414a-ba3f-ea86a7303ac0","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T14:37:09.054160+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"df9e2275-2597-4a3e-8508-aee0717ef17a","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T14:37:09.078211+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8bcace42-911e-43e5-bd4f-3e835685bd4a","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T14:37:09.087794+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"447315cb-071b-42b1-914e-eac8ec018d88","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T14:37:09.087966+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":133.1018749915529,"error_message":null,"error_type":null,"event_id":"9d095021-51fe-4b07-8308-3886f04b12bb","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T14:37:09.221234+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"31ac078d-fb84-4521-8e99-71d827d4aed0","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T14:37:09.228301+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0ecaf564-8d8c-4f50-a662-aa98bf599e8a","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 0, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core\", \"skipped_cases\": 60}","model_key":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T14:37:09.228523+00:00"}]},"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core":{"cells":[{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-06T14:37:06.404488+00:00","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","failure_count":0,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.6833333333333333,"profile":"core","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":6192222208,"deployment":"local","effective_model_id":"qwen3.8-27b-iq1s","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","family":"qwen3.8","model_id":"Qwen/Qwen3.8-27B","model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","parameters_b":27.0,"provider_key":"korgis","quantization":"UD-IQ1_S","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2","model_key":"qwen3.8-27b-iq1s","process_cpu_percent_avg":5.429397449534366,"process_rss_bytes_avg":7678658248.931642,"process_rss_bytes_peak":9314566144.0,"sample_count":1314,"sample_interval_ms":323.2460322738765,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4307910656.0}],"run":{"completed_at_utc":"2026-10-06T14:37:06.404488+00:00","created_at_utc":"2026-10-06T13:42:15.419657+00:00","git_commit":"640138c3429882a42797d28074f98d3fea98a63b","integrity_state":"VALID","manifest_checksum":"48e9324ccb0ab6ce894712ea4854bffbbe5ae4d354755c88013e22cb66781d0c","profile":"core","run_group":"2026-10-core-v1-qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8bc199d7-3e2a-4c43-abca-c93f5def0e87","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"structured-output\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"qwen3.8-27b-iq1s\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T13:42:15.419657+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ccada015-c95f-47bd-9223-e47fcb4d6546","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T13:42:15.456040+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":59.75362501339987,"error_message":null,"error_type":null,"event_id":"abd1baf3-4a1a-4abd-ad62-9c2bf9493c92","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T13:42:15.516144+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2758d121-7ea6-4505-afd0-69bf42166395","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T13:42:15.516379+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"4b1d53a6-2492-4913-8a69-166744e28eba","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T13:42:15.563722+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"38dbca6c-029f-4ec0-b585-f1d76c639a30","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T13:49:04.069826+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"fb74fdcd-6c20-4ff2-bd64-9bb60b6cb93a","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T13:49:04.070699+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":569.0681249980116,"error_message":null,"error_type":null,"event_id":"e76fcf7e-8651-4485-947b-9814905f38bf","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T13:49:04.640002+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"43aee062-594d-4376-b63e-696a5eff3602","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T13:49:04.655521+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"359e9508-43e9-46d5-90e2-a1e885977e23","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T13:49:04.655677+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2a79e061-c0a6-4956-bc5a-5d211908da14","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"structured-output\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"qwen3.8-27b-iq1s\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T14:37:05.695084+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3bb1dab4-7853-459a-a7d2-b0768bafa68e","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T14:37:05.720310+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":50.324040988925844,"error_message":null,"error_type":null,"event_id":"87e56670-7bd2-4251-bba3-a95abaf80e3a","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T14:37:05.771059+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f0ffe925-818a-43d2-938f-b2126b7e4158","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T14:37:05.771305+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b3bad59e-f6ba-4f3b-8474-a6cab52e7066","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T14:37:05.818874+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b0b08556-a7ce-4006-8b7c-c14df653cc82","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T14:37:05.829250+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e83b6508-e9ae-4404-99c3-7f28d9faba8c","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T14:37:05.829410+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":567.4100420001196,"error_message":null,"error_type":null,"event_id":"e0b4096e-7391-4682-958c-1099aa820f59","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 6192222208, \"backend\": \"llama_server\", \"quantization\": \"UD-IQ1_S\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.8-27b-iq1s\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T14:37:06.396992+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"11536641-4c0b-4bf2-8bf8-63b6ef69b622","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:a9c4c74e61ae0adf675f9734cec68dbdb41e4384d7fa4e2d0d1825407fe33fd2\", \"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core\"}","model_key":"qwen3.8-27b-iq1s","run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T14:37:06.404322+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e1efe279-2cd0-4c6f-ac2d-f60b309b6e49","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 0, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"2026-10-core-v1-qwen3.8-27b-iq1s\", \"run_id\": \"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core\", \"skipped_cases\": 60}","model_key":null,"run_id":"2026-10-core-v1-qwen3.8-27b-iq1s-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T14:37:06.404488+00:00"}]},"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core":{"cells":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-06T09:08:01.471590+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6493506493506493,"profile":"core","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"xhtoken/spark-x2.5-4b","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","parameters_b":null,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"spark-x2.5-4b-q4km","process_cpu_percent_avg":5.594450252441077,"process_rss_bytes_avg":3797578739.069748,"process_rss_bytes_peak":6051266560.0,"sample_count":1308,"sample_interval_ms":310.7691591083291,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5891014656.0}],"run":{"completed_at_utc":"2026-10-06T09:08:01.471590+00:00","created_at_utc":"2026-10-06T09:01:46.968894+00:00","git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","integrity_state":"VALID","manifest_checksum":"39f2de8a8497d14c06e3b004ecd329478581cc701e13f3559f5a6bbf5b58bc4f","profile":"core","run_group":"2026-10-core-v1-spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ed64faee-8e6c-4f61-907c-9baafbbb8bcb","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"intent-classification\"], \"model_keys\": [\"spark-x2.5-4b-q4km\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T09:01:46.968894+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"52432df7-375f-4179-aabb-f4f27f0e9577","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:01:46.995556+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2103.1057499931194,"error_message":null,"error_type":null,"event_id":"75670364-c05c-4c1b-a380-cbbab77cf140","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:01:49.102025+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"982a8d10-929d-43d2-9699-9cbf3a6717f5","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:01:49.102500+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"42aeec9a-7f20-4688-bf67-799bb4ea60e4","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:01:49.111851+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"58a912d0-7246-46a3-8ea0-d4b03d106c92","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:08:00.673465+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7b48cc5b-5b3a-48da-be25-9c52e378caa8","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:08:00.674166+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":772.590291002416,"error_message":null,"error_type":null,"event_id":"c1399020-1560-4eb6-b3af-f86b25d8fbd4","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:08:01.446786+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"4c0a8e0b-81fd-4edf-adcd-2a7f53967e5f","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:08:01.471314+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8396554f-7792-4ed3-ab2f-5e70ba30d79a","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T09:08:01.471590+00:00"}]},"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core":{"cells":[{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-06T09:01:45.488881+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.45,"profile":"core","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"xhtoken/spark-x2.5-4b","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","parameters_b":null,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"spark-x2.5-4b-q4km","process_cpu_percent_avg":8.149214468504207,"process_rss_bytes_avg":3498299289.6,"process_rss_bytes_peak":3887923200.0,"sample_count":147,"sample_interval_ms":267.7047739123735,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":7349649408.0}],"run":{"completed_at_utc":"2026-10-06T09:01:45.488881+00:00","created_at_utc":"2026-10-06T09:01:11.533005+00:00","git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","integrity_state":"VALID","manifest_checksum":"5670fc30a958ff61ee03867fca49bcc8092d8975451026202c1962874696bf11","profile":"core","run_group":"2026-10-core-v1-spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ce63d657-61a9-42ed-aa27-90ccd149f290","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"model_keys\": [\"spark-x2.5-4b-q4km\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T09:01:11.533005+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"abad6d3c-8314-4414-a317-6750f250ad16","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:01:11.567935+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2111.6234159999294,"error_message":null,"error_type":null,"event_id":"316c8be9-15b2-4cc9-9708-f3a0858585eb","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:01:13.680499+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d1be0eca-d1ad-48a3-a5a9-eaab0f2aab5e","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:01:13.680868+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"207f15be-aebf-4961-8277-e437216c7ec6","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:01:13.706969+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"4e029f9a-cbd6-49ea-a839-7e08b32f84d2","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:01:45.194222+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c9032287-eb2b-4af7-b8a7-464faaa307f1","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:01:45.194465+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":255.08633299614303,"error_message":null,"error_type":null,"event_id":"708b6f24-4891-4ea6-84ef-a57dcc044e91","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:01:45.449728+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c8951f01-b976-4837-81c1-cd5b5dcefebc","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:01:45.474219+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"063377d4-659d-45e8-a148-ba70de7a8ef6","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 40, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 40, \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T09:01:45.488881+00:00"}]},"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core":{"cells":[{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-06T09:14:39.475593+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"profile":"core","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"xhtoken/spark-x2.5-4b","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","parameters_b":null,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"spark-x2.5-4b-q4km","process_cpu_percent_avg":5.353823389553569,"process_rss_bytes_avg":4696997069.724959,"process_rss_bytes_peak":7862026240.0,"sample_count":1416,"sample_interval_ms":302.2578173696654,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6040551424.0}],"run":{"completed_at_utc":"2026-10-06T09:14:39.475593+00:00","created_at_utc":"2026-10-06T09:08:02.957031+00:00","git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","integrity_state":"VALID","manifest_checksum":"77df49d6f006fe5321f6c2f66a5aec350606167e9f6487c2a25a9919745e1fbc","profile":"core","run_group":"2026-10-core-v1-spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0c36fbe1-5c9e-4d3d-a9cc-e3220f96d197","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"oos-calibration\"], \"model_keys\": [\"spark-x2.5-4b-q4km\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T09:08:02.957031+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"5f3381b5-8afd-4ab1-8e91-68467dc16f46","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:08:02.982508+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2106.969417000073,"error_message":null,"error_type":null,"event_id":"4da5109b-55b4-40bd-b814-1d4e1b0d6e5b","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:08:05.090224+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"78b5a969-81ab-4979-ad06-8d34c539838e","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:08:05.090626+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"52897cee-e0c7-423e-81d0-63aa45ef1811","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:08:05.122558+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"93b96f0c-3c6f-43cb-8e50-fd273d78d3f0","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:14:38.904148+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"de242cf4-e561-4b98-804c-936ab9606f1b","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:14:38.904391+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":558.9563750108937,"error_message":null,"error_type":null,"event_id":"e1b64e83-d2da-49f1-8faf-854b7feb7b67","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:14:39.463529+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"17439409-ae78-43ff-a907-87d340e7fdf7","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:14:39.475296+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"07a04242-99bf-428c-b818-dad84c3db3ed","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T09:14:39.475593+00:00"}]},"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core":{"cells":[{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-06T09:01:09.715786+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.95,"profile":"core","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"xhtoken/spark-x2.5-4b","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","parameters_b":null,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"spark-x2.5-4b-q4km","process_cpu_percent_avg":9.07249993238987,"process_rss_bytes_avg":3570442922.666667,"process_rss_bytes_peak":4295933952.0,"sample_count":244,"sample_interval_ms":280.7728714734669,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6034112512.0}],"run":{"completed_at_utc":"2026-10-06T09:01:09.715786+00:00","created_at_utc":"2026-10-06T09:00:09.719791+00:00","git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","integrity_state":"VALID","manifest_checksum":"12b55c7717253c7067b93ae7991d3cbaeaafca39c6437e5ac5cd0be154c2069c","profile":"core","run_group":"2026-10-core-v1-spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b8a5b637-602a-4a2e-9312-c967433614db","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"qa-abstention\"], \"model_keys\": [\"spark-x2.5-4b-q4km\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T09:00:09.719791+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"508d1670-af82-416d-8d80-70a40c4d281e","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:00:09.747854+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2121.2086669984274,"error_message":null,"error_type":null,"event_id":"6093f925-6f39-49eb-aa60-954e1217c3a8","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:00:11.869724+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"003aaba2-fd38-40db-9d21-78203d574601","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:00:11.870027+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"9fe89f72-9174-4c8a-9357-fdab0b47a1b0","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:00:11.929998+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f69fa20d-c855-4f85-9c00-df5715a53322","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:01:09.299050+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7d04dc56-e931-4ad9-aee6-32a182069204","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:01:09.300282+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":401.4078750042245,"error_message":null,"error_type":null,"event_id":"6f65f1d1-6be9-4bd0-bf9b-5c836c878dc3","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:01:09.702389+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d1c1db87-e4e8-4f50-b849-65340847e9bc","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:01:09.715226+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"81f16cdf-79f9-438f-bf18-627ce47cfcc2","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T09:01:09.715786+00:00"}]},"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core":{"cells":[{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-06T09:00:08.668486+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7666666666666667,"profile":"core","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"xhtoken/spark-x2.5-4b","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944","parameters_b":null,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"spark-x2.5-4b-q4km","process_cpu_percent_avg":7.982890411325502,"process_rss_bytes_avg":3362284766.8271284,"process_rss_bytes_peak":3593027584.0,"sample_count":417,"sample_interval_ms":298.61884293800324,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5815320576.0}],"run":{"completed_at_utc":"2026-10-06T09:00:08.668486+00:00","created_at_utc":"2026-10-06T08:48:05.026982+00:00","git_commit":"08bbb92824f3070d93fb3c16b5847b46dec03dfb","integrity_state":"VALID","manifest_checksum":"4eb6f80a0fdca445f8a0eb8466447cb33451bbd32d781ec6643aafacbd95d317","profile":"core","run_group":"2026-10-core-v1-spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7b3ff508-1359-42b3-97ee-0ffd569f897b","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"structured-output\"], \"model_keys\": [\"spark-x2.5-4b-q4km\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T08:48:05.026982+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"12a5a010-ae01-47b1-aca3-ed06453876fe","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T08:48:05.060190+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"54f8ee2f-fbff-47f1-9a35-52c06b3ba1dd","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"structured-output\"], \"model_keys\": [\"spark-x2.5-4b-q4km\"], \"profile\": \"core\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T08:58:16.686692+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2810447e-7fb3-475b-82fa-8f2ec69542fa","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T08:58:16.711648+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":74.79937499738298,"error_message":null,"error_type":null,"event_id":"b46869aa-6ae8-4083-bc73-c4912cefd0c1","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T08:58:16.787428+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"68057b61-c4e7-4c69-b708-63a68692c343","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T08:58:16.787855+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"bc2e239e-751e-44da-9046-abd54b302c93","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T08:58:16.856862+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2d5ba325-2ab3-4272-8277-94a247a40f71","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:00:08.363530+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"68b685d4-1ea3-4d47-b1ae-ca68fc417ea7","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-06T09:00:08.363809+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":294.78116700192913,"error_message":null,"error_type":null,"event_id":"33a2bf72-e8c2-4123-b9a5-b386c77508d1","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:00:08.658801+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"5248c0fc-1a46-4130-9d8f-652086476720","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:a7eeca234da304612647bed0c5f28780b93685c7dcec0442bc95393a328ea944\", \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core\"}","model_key":"spark-x2.5-4b-q4km","run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-06T09:00:08.668218+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2e05bff8-9735-423b-882c-a67c7c924f79","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"2026-10-core-v1-spark-x2.5-4b-q4km\", \"run_id\": \"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-core-v1-spark-x2.5-4b-q4km-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-06T09:00:08.668486+00:00"}]},"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-intent-classification-core":{"cells":[{"benchmark_signature":"sha256:benchmark:d6ab2ffb1f34c76d5dddef81996eb87dc7f2d20e82b7d6e34eff857d6c17bb72","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-22T09:44:42.382206+00:00","execution_signature":"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6493506493506493,"profile":"core","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-intent-classification-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":10947100928,"deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_signature":"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa","family":"lfm2.5","model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"provider_key":"korgis","quantization":"UD-Q3_K_XL","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-intent-classification-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","process_cpu_percent_avg":9.709798039457981,"process_rss_bytes_avg":11220907372.332468,"process_rss_bytes_peak":11379949568.0,"sample_count":10139,"sample_interval_ms":317.0364510559339,"sampling_error_count":1,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":2392195072.0}],"run":{"completed_at_utc":"2026-10-22T09:44:42.382206+00:00","created_at_utc":"2026-10-22T08:51:39.298035+00:00","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","integrity_state":"VALID","manifest_checksum":"fd4427541a214849a942e2351a971ffa752ef9d68698969ea1d535e219ca67fb","profile":"core","run_group":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-intent-classification-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"3"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7c2c40df-d1ec-4dff-88cd-52756997c2e0","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"intent-classification\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"lfm2.5-24b-a2b-ud-q3-k-xl\"], \"profile\": \"core\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-intent-classification-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"3\"}","model_key":null,"run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-22T08:51:39.298035+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"fd2dc939-08e9-4652-8f41-3ec43f6f0ae1","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T08:51:39.319046+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2198.198250029236,"error_message":null,"error_type":null,"event_id":"a70422bc-6037-461e-a3a5-373aa8691c9d","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T08:51:41.517681+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3f14eebf-dbb4-4466-8146-5c435c449aae","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T08:51:41.517969+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"da7490b1-0ecc-4cd6-aedd-191a8a342871","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:d6ab2ffb1f34c76d5dddef81996eb87dc7f2d20e82b7d6e34eff857d6c17bb72\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T08:51:41.525019+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"da6ed5b3-e108-4625-9ba0-4edee002d844","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:d6ab2ffb1f34c76d5dddef81996eb87dc7f2d20e82b7d6e34eff857d6c17bb72\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T09:44:42.006887+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"4a46f8b1-ee11-43fe-80a9-cdbadd6fff3b","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T09:44:42.007527+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":351.44766594748944,"error_message":null,"error_type":null,"event_id":"7fbee3d5-0c93-4ad0-bd16-4a7080e27c8a","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T09:44:42.359245+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0511254a-2d5b-4297-84d3-02ac3ba66433","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:d6ab2ffb1f34c76d5dddef81996eb87dc7f2d20e82b7d6e34eff857d6c17bb72\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-intent-classification-core\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T09:44:42.381774+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"4dcf5680-8081-4021-accf-26dc9a7384c6","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-intent-classification-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-22T09:44:42.382206+00:00"}]},"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-mathematical-reasoning-core":{"cells":[{"benchmark_signature":"sha256:benchmark:a5db4a5d2c68fabba1fda0494f04219381d92bf682959ba51aed96f18f3605d2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-22T08:51:38.368018+00:00","execution_signature":"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.95,"profile":"core","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-mathematical-reasoning-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":10947100928,"deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_signature":"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa","family":"lfm2.5","model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"provider_key":"korgis","quantization":"UD-Q3_K_XL","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-mathematical-reasoning-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","process_cpu_percent_avg":9.76682190243444,"process_rss_bytes_avg":10495045714.774097,"process_rss_bytes_peak":10535059456.0,"sample_count":1221,"sample_interval_ms":311.0348310753173,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4095033344.0}],"run":{"completed_at_utc":"2026-10-22T08:51:38.368018+00:00","created_at_utc":"2026-10-22T08:45:23.526981+00:00","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","integrity_state":"VALID","manifest_checksum":"6f0940e0a7d57020261afd6184bbdc2f72674291a065b58ecf6255ff891c5fe4","profile":"core","run_group":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-mathematical-reasoning-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"3"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7f2e1da1-a78a-43b0-80f1-9a0143853f09","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"lfm2.5-24b-a2b-ud-q3-k-xl\"], \"profile\": \"core\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-mathematical-reasoning-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"3\"}","model_key":null,"run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-22T08:45:23.526981+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c2bda3b4-a521-4e00-810e-83e7e81d865a","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T08:45:23.540541+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":4108.258791966364,"error_message":null,"error_type":null,"event_id":"099d593b-6334-4eda-92ab-51557aa25257","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T08:45:27.649182+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"9fac5ed2-24f9-4e21-9dca-59c06cf2f84e","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T08:45:27.649481+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"52710077-4d95-4d01-a58d-561872acdc3f","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:a5db4a5d2c68fabba1fda0494f04219381d92bf682959ba51aed96f18f3605d2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T08:45:27.669491+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"cb28d67a-dddc-4dc7-aa8f-c75e5cc688b5","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:a5db4a5d2c68fabba1fda0494f04219381d92bf682959ba51aed96f18f3605d2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T08:51:38.130180+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"9022af20-968c-424a-9322-f7dc824c01f5","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T08:51:38.130365+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":231.57129203900695,"error_message":null,"error_type":null,"event_id":"9ed3da54-caf9-4c18-9aef-6c9544cef414","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T08:51:38.362181+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"51849352-eec8-40c6-9bd9-8cfb97a6d96d","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:a5db4a5d2c68fabba1fda0494f04219381d92bf682959ba51aed96f18f3605d2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-mathematical-reasoning-core\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T08:51:38.367810+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"312ef0ae-05ac-4b07-8c40-67ba2ddf7d9e","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 40, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 40, \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-mathematical-reasoning-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-22T08:51:38.368018+00:00"}]},"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-oos-calibration-core":{"cells":[{"benchmark_signature":"sha256:benchmark:230c029eec827b8bb748e50afe4cc525fae176746626e1dc4cb43e9743332e65","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-22T10:24:06.430109+00:00","execution_signature":"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.974025974025974,"profile":"core","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-oos-calibration-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":10947100928,"deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_signature":"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa","family":"lfm2.5","model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"provider_key":"korgis","quantization":"UD-Q3_K_XL","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-oos-calibration-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","process_cpu_percent_avg":9.992341870095668,"process_rss_bytes_avg":11253426984.76044,"process_rss_bytes_peak":11408441344.0,"sample_count":7601,"sample_interval_ms":314.92489388376083,"sampling_error_count":1,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4138057728.0}],"run":{"completed_at_utc":"2026-10-22T10:24:06.430109+00:00","created_at_utc":"2026-10-22T09:44:43.399412+00:00","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","integrity_state":"VALID","manifest_checksum":"adc478a4cc731d6468685075bb4fdc113e523a9d765a3209989619edeee45a61","profile":"core","run_group":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-oos-calibration-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"3"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"09d33f33-2ab7-4942-8e6d-7c28fee01273","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"oos-calibration\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"lfm2.5-24b-a2b-ud-q3-k-xl\"], \"profile\": \"core\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-oos-calibration-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"3\"}","model_key":null,"run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-22T09:44:43.399412+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3d75c730-5ca1-4aa4-ba1a-6cd9b26fe6d1","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T09:44:43.421951+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2104.973915964365,"error_message":null,"error_type":null,"event_id":"e1cfa5de-87a7-4d06-be13-8532596507aa","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T09:44:45.527370+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f08ff939-c38c-4b61-8fe7-69c177e3536a","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T09:44:45.527609+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"fb245f9a-4b80-4ddc-9fb1-d0599c9b0f7c","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:230c029eec827b8bb748e50afe4cc525fae176746626e1dc4cb43e9743332e65\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T09:44:45.546876+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"9041cb8c-df2d-474a-bed6-4d96c09411fd","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:230c029eec827b8bb748e50afe4cc525fae176746626e1dc4cb43e9743332e65\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T10:24:05.930729+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"644d8c1e-8e06-4edd-93fc-3c5edf705d82","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T10:24:05.931340+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":453.1753329792991,"error_message":null,"error_type":null,"event_id":"595d0d69-7a6c-4b20-a1be-48e541d2622c","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T10:24:06.384791+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0fc4eafc-9862-4414-bf3a-680e71ed8296","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:230c029eec827b8bb748e50afe4cc525fae176746626e1dc4cb43e9743332e65\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-oos-calibration-core\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T10:24:06.429912+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"63c6feab-9909-4df1-9056-4bdc0b1826da","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-oos-calibration-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-22T10:24:06.430109+00:00"}]},"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-qa-abstention-core":{"cells":[{"benchmark_signature":"sha256:benchmark:5ef83b758c4b66715eca88f843e6e438a1f602661374de3eac156fc10ef9f203","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-22T08:45:22.525820+00:00","execution_signature":"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"profile":"core","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-qa-abstention-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":10947100928,"deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_signature":"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa","family":"lfm2.5","model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"provider_key":"korgis","quantization":"UD-Q3_K_XL","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-qa-abstention-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","process_cpu_percent_avg":14.637675854130892,"process_rss_bytes_avg":11191669512.30868,"process_rss_bytes_peak":11286167552.0,"sample_count":1503,"sample_interval_ms":381.115703580942,"sampling_error_count":9,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4156391424.0}],"run":{"completed_at_utc":"2026-10-22T08:45:22.525820+00:00","created_at_utc":"2026-10-22T08:16:15.433024+00:00","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","integrity_state":"VALID","manifest_checksum":"48bff421e9dafe622477d0662b2c0716ae488c39a0aad0140a4d4c6043f97834","profile":"core","run_group":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-qa-abstention-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"3"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"01439383-4f68-450d-833e-ecb6b3644fb2","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"qa-abstention\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"lfm2.5-24b-a2b-ud-q3-k-xl\"], \"profile\": \"core\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-qa-abstention-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"3\"}","model_key":null,"run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-22T08:16:15.433024+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0ab3d86f-d7c7-4540-9831-90fbac7b5388","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T08:16:15.452251+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2114.373499993235,"error_message":null,"error_type":null,"event_id":"f0cb6024-d22b-41bf-b692-0fc31790d43b","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T08:16:17.567133+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"abb7a4f5-8f23-43b1-b75d-d31184f5eb20","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T08:16:17.567426+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f12cbe8e-5510-46b7-8b54-cba3baa52869","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:5ef83b758c4b66715eca88f843e6e438a1f602661374de3eac156fc10ef9f203\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T08:16:17.598655+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"9aca13e4-23de-4705-aa2b-07dfda7fc264","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:5ef83b758c4b66715eca88f843e6e438a1f602661374de3eac156fc10ef9f203\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T08:45:22.084818+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a9aa934d-c168-44c8-8c80-5f0842380203","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T08:45:22.085040+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":417.8599170409143,"error_message":null,"error_type":null,"event_id":"e9b29056-5d32-4626-84fd-b6bf89b73c21","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T08:45:22.503053+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7aebd5b2-49ed-493c-a651-62ca03352933","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:5ef83b758c4b66715eca88f843e6e438a1f602661374de3eac156fc10ef9f203\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-qa-abstention-core\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T08:45:22.525120+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"01fd8d4a-9329-439c-acc3-8af489b4b096","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-qa-abstention-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-22T08:45:22.525820+00:00"}]},"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core":{"cells":[{"benchmark_signature":"sha256:benchmark:b8edb9ae85ac7482f2930cdbf8920bc573cea8dded0851e3351b0e05a8900d88","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-22T08:16:14.436917+00:00","execution_signature":"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7666666666666667,"profile":"core","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core","runtime_json":"{}","sweep_id":null},"failure_summary":[{"count":1,"error_type":"TransportError","event_type":"model.release.failed","first_at":"2026-10-22T07:57:01.895799+00:00","last_at":"2026-10-22T07:57:01.895799+00:00"}],"models":[{"artifact_format":"gguf","artifact_size_bytes":10947100928,"deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_signature":"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa","family":"lfm2.5","model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"provider_key":"korgis","quantization":"UD-Q3_K_XL","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","process_cpu_percent_avg":16.534905097118784,"process_rss_bytes_avg":2426066860.6895175,"process_rss_bytes_peak":8312995840.0,"sample_count":2135,"sample_interval_ms":327.97111656558457,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":3762388992.0}],"run":{"completed_at_utc":"2026-10-22T08:16:14.436917+00:00","created_at_utc":"2026-10-22T07:56:31.502889+00:00","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","integrity_state":"VALID","manifest_checksum":"0cbce4ede0dba62fed1c64baa483a0c39a98e82608cb00881617dd1332fff6f5","profile":"core","run_group":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"3"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"5918e135-4965-4529-839a-f8d2e367e0f8","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"structured-output\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"lfm2.5-24b-a2b-ud-q3-k-xl\"], \"profile\": \"core\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"3\"}","model_key":null,"run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-22T07:56:31.502889+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"77584ee5-71f4-476f-9cd0-e35e4700601a","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T07:56:31.532816+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":57.90949997026473,"error_message":null,"error_type":null,"event_id":"5f2349e5-2121-409c-9e77-9cc314ce6e67","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T07:56:31.591232+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"bfdcaed7-078f-46c6-8166-7f2d2d32642b","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T07:56:31.591560+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e99c90b6-8b4f-4107-9176-15d9db7c02f2","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:833a641c565eb63fd8c740f967972ebbbd7789d8c7dbf8fe6b81ee5d9a57ea48\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T07:56:31.659595+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"efd54924-2263-422b-ac0e-eadf33d4f398","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T07:57:01.893055+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":"HTTP 409: {\"detail\":\"Model 'lfm2.5-24b-a2b-ud-q3-k-xl' has an active request.\"}","error_type":"TransportError","event_id":"c7883888-5eef-4113-b4f9-ab9227edf62f","event_type":"model.release.failed","metadata_json":"{\"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core","sample_id":null,"status":"failed","timestamp_utc":"2026-10-22T07:57:01.895799+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c12c31ca-dce6-4b5e-8ce5-e62b042b7227","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"structured-output\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"lfm2.5-24b-a2b-ud-q3-k-xl\"], \"profile\": \"core\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"3\"}","model_key":null,"run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-22T08:05:10.574228+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e422762b-5187-4485-a973-34d1f1fc3d42","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T08:05:10.601544+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":75.79279190395027,"error_message":null,"error_type":null,"event_id":"e9651848-58e8-4633-910f-32936f4827a7","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T08:05:10.677941+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"634a9e01-ed7f-4e39-9a77-29752c478c76","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T08:05:10.678268+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f2bcc13f-90a0-4e0f-81e9-29899d665ddf","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:b8edb9ae85ac7482f2930cdbf8920bc573cea8dded0851e3351b0e05a8900d88\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T08:05:10.745738+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"901c4e66-2473-417c-aa3a-f83101b48aa4","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:b8edb9ae85ac7482f2930cdbf8920bc573cea8dded0851e3351b0e05a8900d88\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T08:16:13.892464+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0508e5e2-8ecb-46b4-91d8-be58f794bbdc","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T08:16:13.892715+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":512.3935410520062,"error_message":null,"error_type":null,"event_id":"46454cb1-067c-48ee-b8a4-4211cb4310da","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T08:16:14.405318+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e16a32d9-b1f0-43d9-aeab-00b8c6ea16c9","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:b8edb9ae85ac7482f2930cdbf8920bc573cea8dded0851e3351b0e05a8900d88\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:f25213284731d4f47331096f68af4dfd7d4244602f5a37d535de1303416a4caa\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T08:16:14.436700+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d366bd16-82ff-4add-90c8-b7c8506720aa","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl\", \"run_id\": \"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"2026-10-lfm25-core-v3-lfm2.5-24b-a2b-ud-q3-k-xl-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-22T08:16:14.436917+00:00"}]},"2533bf73-b012-4d12-a6b6-9b6e4334dae2":{"cells":[{"benchmark_signature":"sha256:benchmark:382be54f0e8c250212fa1d53656a34eb22335c491d7aa5b1ba59e927a0c58dea","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T14:59:31.170321+00:00","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","failure_count":0,"model_key":"limite-1b-violetto","model_signature":"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.7222222222222222,"profile":"extended","run_id":"2533bf73-b012-4d12-a6b6-9b6e4334dae2","sample_count":72,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"2533bf73-b012-4d12-a6b6-9b6e4334dae2","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","family":"limite","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65","parameters_b":1.0,"provider_key":"transformers","quantization":null,"run_id":"2533bf73-b012-4d12-a6b6-9b6e4334dae2","runtime_key":"transformers-local"}],"resources":[],"run":{"completed_at_utc":"2026-10-15T14:59:31.170321+00:00","created_at_utc":"2026-10-15T14:46:26.133597+00:00","git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","integrity_state":"VALID","manifest_checksum":"9c078767f485c35e931045ebb32e59b5614647c25410d7b2b3db8ba5a7983412","profile":"extended","run_group":"math-v3-extended","run_id":"2533bf73-b012-4d12-a6b6-9b6e4334dae2","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"3"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7ad4d35a-5afe-4e00-9bf0-aa845249c7dc","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"limite-1b-violetto\"], \"profile\": \"extended\", \"run_group\": \"math-v3-extended\", \"run_id\": \"2533bf73-b012-4d12-a6b6-9b6e4334dae2\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"3\"}","model_key":null,"run_id":"2533bf73-b012-4d12-a6b6-9b6e4334dae2","sample_id":null,"status":null,"timestamp_utc":"2026-10-15T14:46:26.133597+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"326cde91-c142-4be7-b9de-c030a95fc7fe","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65\", \"provider_key\": \"transformers\", \"run_group\": \"math-v3-extended\", \"run_id\": \"2533bf73-b012-4d12-a6b6-9b6e4334dae2\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"2533bf73-b012-4d12-a6b6-9b6e4334dae2","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T14:46:26.149702+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.2050419570878148,"error_message":null,"error_type":null,"event_id":"ce85ab48-b531-4abb-8d0e-b36193d37362","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65\", \"provider_key\": \"transformers\", \"run_group\": \"math-v3-extended\", \"run_id\": \"2533bf73-b012-4d12-a6b6-9b6e4334dae2\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"2533bf73-b012-4d12-a6b6-9b6e4334dae2","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T14:46:26.150250+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"75ef8abf-ea8d-4c28-8223-7a025a82fcf5","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65\", \"provider_key\": \"transformers\", \"run_group\": \"math-v3-extended\", \"run_id\": \"2533bf73-b012-4d12-a6b6-9b6e4334dae2\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"2533bf73-b012-4d12-a6b6-9b6e4334dae2","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T14:46:26.150396+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e83cbc71-13b6-46c9-a48b-3970ca18480e","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:382be54f0e8c250212fa1d53656a34eb22335c491d7aa5b1ba59e927a0c58dea\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65\", \"planned_cases\": 72, \"profile\": \"extended\", \"provider_key\": \"transformers\", \"run_group\": \"math-v3-extended\", \"run_id\": \"2533bf73-b012-4d12-a6b6-9b6e4334dae2\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"2533bf73-b012-4d12-a6b6-9b6e4334dae2","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T14:46:26.166748+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"af65dafd-58cd-4ad6-a67f-d872ca65415a","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:382be54f0e8c250212fa1d53656a34eb22335c491d7aa5b1ba59e927a0c58dea\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65\", \"profile\": \"extended\", \"provider_key\": \"transformers\", \"run_group\": \"math-v3-extended\", \"run_id\": \"2533bf73-b012-4d12-a6b6-9b6e4334dae2\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"2533bf73-b012-4d12-a6b6-9b6e4334dae2","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T14:59:30.838877+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"bca22736-b07d-40c9-91d7-086004484fb7","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65\", \"provider_key\": \"transformers\", \"run_group\": \"math-v3-extended\", \"run_id\": \"2533bf73-b012-4d12-a6b6-9b6e4334dae2\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"2533bf73-b012-4d12-a6b6-9b6e4334dae2","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T14:59:30.839200+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":324.25091601908207,"error_message":null,"error_type":null,"event_id":"0683ac3b-d997-430d-a22f-bf8765025247","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65\", \"provider_key\": \"transformers\", \"run_group\": \"math-v3-extended\", \"run_id\": \"2533bf73-b012-4d12-a6b6-9b6e4334dae2\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"2533bf73-b012-4d12-a6b6-9b6e4334dae2","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T14:59:31.163691+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ca068c48-7423-462e-9aff-790669cf269a","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:382be54f0e8c250212fa1d53656a34eb22335c491d7aa5b1ba59e927a0c58dea\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65\", \"run_group\": \"math-v3-extended\", \"run_id\": \"2533bf73-b012-4d12-a6b6-9b6e4334dae2\"}","model_key":"limite-1b-violetto","run_id":"2533bf73-b012-4d12-a6b6-9b6e4334dae2","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T14:59:31.170161+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"bb4d6be4-1995-4f5a-86c0-a3a7e69126d9","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 72, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 72, \"run_group\": \"math-v3-extended\", \"run_id\": \"2533bf73-b012-4d12-a6b6-9b6e4334dae2\", \"skipped_cases\": 0}","model_key":null,"run_id":"2533bf73-b012-4d12-a6b6-9b6e4334dae2","sample_id":null,"status":null,"timestamp_utc":"2026-10-15T14:59:31.170321+00:00"}]},"3a46714c-d579-4688-a8b0-558183b4e02a":{"cells":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T16:36:39.120802+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6688311688311688,"profile":"core","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"},{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-05T16:36:39.120802+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.475,"profile":"core","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-05T16:36:39.120802+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.974025974025974,"profile":"core","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"},{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-05T16:36:39.120802+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"profile":"core","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"},{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-05T16:36:39.120802+00:00","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","failure_count":0,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7333333333333333,"profile":"core","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","family":null,"model_id":"unsloth/qwen3.5-4b","model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","parameters_b":null,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-4b-q4km","process_cpu_percent_avg":6.008969494809468,"process_rss_bytes_avg":5011243553.1201,"process_rss_bytes_peak":7911129088.0,"sample_count":1067,"sample_interval_ms":281.19401557671836,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5343313920.0},{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-4b-q4km","process_cpu_percent_avg":12.689180073192913,"process_rss_bytes_avg":5970077900.800001,"process_rss_bytes_peak":7198195712.0,"sample_count":134,"sample_interval_ms":272.2716644006747,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":8184053760.0},{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-4b-q4km","process_cpu_percent_avg":6.406836413976952,"process_rss_bytes_avg":6021487127.958771,"process_rss_bytes_peak":10610851840.0,"sample_count":1193,"sample_interval_ms":286.7970046134617,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5619220480.0},{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-4b-q4km","process_cpu_percent_avg":9.78397589283007,"process_rss_bytes_avg":7517191058.123175,"process_rss_bytes_peak":10744905728.0,"sample_count":261,"sample_interval_ms":276.8212083626713,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":6344114176.0},{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b","model_key":"qwen3.5-4b-q4km","process_cpu_percent_avg":8.486342485499302,"process_rss_bytes_avg":5121284030.324117,"process_rss_bytes_peak":6889406464.0,"sample_count":411,"sample_interval_ms":286.355363495796,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5873369088.0}],"run":{"completed_at_utc":"2026-10-05T16:36:39.120802+00:00","created_at_utc":"2026-10-05T16:23:54.056104+00:00","git_commit":"cb061291446b8dbd79f524a4fefcd4a4406e2a0d","integrity_state":"VALID","manifest_checksum":"e32e59488bed055fe404dd69e3f7e7ec6ff4f37abb1884b4164dfa73d6335cb7","profile":"core","run_group":"qwen3.5-4b-core-all","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6bfaf962-48ba-473f-8c07-c2973d8cc129","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"intent-classification\", \"oos-calibration\", \"structured-output\", \"qa-abstention\", \"mathematical-reasoning\"], \"model_keys\": [\"qwen3.5-4b-q4km\"], \"profile\": \"core\", \"run_group\": \"qwen3.5-4b-core-all\", \"run_id\": \"3a46714c-d579-4688-a8b0-558183b4e02a\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T16:23:54.056104+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"569b751c-aadc-431c-93d1-1542697724ad","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"provider_key\": \"korgis\", \"run_group\": \"qwen3.5-4b-core-all\", \"run_id\": \"3a46714c-d579-4688-a8b0-558183b4e02a\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q4km","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T16:23:54.077172+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2095.40270800062,"error_message":null,"error_type":null,"event_id":"6881635b-56fe-4db7-b920-4d9277343b22","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"provider_key\": \"korgis\", \"run_group\": \"qwen3.5-4b-core-all\", \"run_id\": \"3a46714c-d579-4688-a8b0-558183b4e02a\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q4km","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T16:23:56.173113+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"599ae35f-5710-41b3-8562-cf454a4a1df5","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"provider_key\": \"korgis\", \"run_group\": \"qwen3.5-4b-core-all\", \"run_id\": \"3a46714c-d579-4688-a8b0-558183b4e02a\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q4km","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T16:23:56.173408+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"594a8a32-c013-4ac2-a401-a83eac9000ef","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen3.5-4b-core-all\", \"run_id\": \"3a46714c-d579-4688-a8b0-558183b4e02a\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q4km","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T16:23:56.181224+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8b7d3251-b0c1-489c-bab6-68317f7f39f6","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen3.5-4b-core-all\", \"run_id\": \"3a46714c-d579-4688-a8b0-558183b4e02a\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q4km","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T16:28:19.983284+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f51cb432-0963-4551-89df-0e10bf5c8699","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"provider_key\": \"korgis\", \"run_group\": \"qwen3.5-4b-core-all\", \"run_id\": \"3a46714c-d579-4688-a8b0-558183b4e02a\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q4km","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T16:28:19.983474+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"bac529c7-7b40-41e1-9326-ca5fadf27aaf","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen3.5-4b-core-all\", \"run_id\": \"3a46714c-d579-4688-a8b0-558183b4e02a\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q4km","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T16:28:20.003861+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ef68ad2e-c82b-436e-8c60-74eddf69d30a","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen3.5-4b-core-all\", \"run_id\": \"3a46714c-d579-4688-a8b0-558183b4e02a\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q4km","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T16:33:26.051048+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1dec8d9d-c28b-44a9-8789-ef8f3bc6107d","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"provider_key\": \"korgis\", \"run_group\": \"qwen3.5-4b-core-all\", \"run_id\": \"3a46714c-d579-4688-a8b0-558183b4e02a\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q4km","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T16:33:26.051246+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7a940640-e7ea-4cfd-9120-c54d8376cd1e","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen3.5-4b-core-all\", \"run_id\": \"3a46714c-d579-4688-a8b0-558183b4e02a\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q4km","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T16:33:26.129040+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e8e5d2ce-a7d1-4290-8ff2-0c1a627cfb1c","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen3.5-4b-core-all\", \"run_id\": \"3a46714c-d579-4688-a8b0-558183b4e02a\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q4km","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T16:35:11.432530+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"752af077-d57d-4a61-a93b-04c49e54f9e0","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"provider_key\": \"korgis\", \"run_group\": \"qwen3.5-4b-core-all\", \"run_id\": \"3a46714c-d579-4688-a8b0-558183b4e02a\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q4km","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T16:35:11.432776+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b8993386-3725-4326-bd25-1ab94acf485a","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen3.5-4b-core-all\", \"run_id\": \"3a46714c-d579-4688-a8b0-558183b4e02a\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q4km","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T16:35:11.459044+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"444f12d4-dea8-4abe-a805-3435ab658470","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen3.5-4b-core-all\", \"run_id\": \"3a46714c-d579-4688-a8b0-558183b4e02a\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q4km","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T16:36:11.014172+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0d4933f1-03e5-4952-8a57-f1834b208cb2","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"provider_key\": \"korgis\", \"run_group\": \"qwen3.5-4b-core-all\", \"run_id\": \"3a46714c-d579-4688-a8b0-558183b4e02a\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q4km","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T16:36:11.014403+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f3690cf6-a588-4f0a-b507-0804dbeef4de","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen3.5-4b-core-all\", \"run_id\": \"3a46714c-d579-4688-a8b0-558183b4e02a\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q4km","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T16:36:11.028477+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3d51671a-c42d-4fc0-b5c2-429ce7194caf","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen3.5-4b-core-all\", \"run_id\": \"3a46714c-d579-4688-a8b0-558183b4e02a\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q4km","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T16:36:38.653125+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8807a704-242d-40f1-981b-f802e887df86","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"provider_key\": \"korgis\", \"run_group\": \"qwen3.5-4b-core-all\", \"run_id\": \"3a46714c-d579-4688-a8b0-558183b4e02a\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q4km","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T16:36:38.653346+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":440.4519590025302,"error_message":null,"error_type":null,"event_id":"57530c3d-7a2a-4ebd-8989-ebe20405e551","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"provider_key\": \"korgis\", \"run_group\": \"qwen3.5-4b-core-all\", \"run_id\": \"3a46714c-d579-4688-a8b0-558183b4e02a\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q4km","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T16:36:39.093979+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a7640e3a-b475-4be9-ac75-ca89bcf35d91","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"run_group\": \"qwen3.5-4b-core-all\", \"run_id\": \"3a46714c-d579-4688-a8b0-558183b4e02a\"}","model_key":"qwen3.5-4b-q4km","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T16:36:39.114523+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c6bb92c7-bc17-4497-b4a6-1d4788f367c2","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"run_group\": \"qwen3.5-4b-core-all\", \"run_id\": \"3a46714c-d579-4688-a8b0-558183b4e02a\"}","model_key":"qwen3.5-4b-q4km","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T16:36:39.116586+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c5a6f133-d173-4e49-8ce3-cdaa66e76730","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"run_group\": \"qwen3.5-4b-core-all\", \"run_id\": \"3a46714c-d579-4688-a8b0-558183b4e02a\"}","model_key":"qwen3.5-4b-q4km","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T16:36:39.118110+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"53951052-01a0-49c3-acd3-7a1e57d84b8f","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"run_group\": \"qwen3.5-4b-core-all\", \"run_id\": \"3a46714c-d579-4688-a8b0-558183b4e02a\"}","model_key":"qwen3.5-4b-q4km","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T16:36:39.119478+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"bf0b01ee-0d45-44dc-b203-81a90563983a","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:7af693b03b87a55605b1c7c43daf0f7e8b6e2ab607b934628d5d633f8f38f67b\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"run_group\": \"qwen3.5-4b-core-all\", \"run_id\": \"3a46714c-d579-4688-a8b0-558183b4e02a\"}","model_key":"qwen3.5-4b-q4km","run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T16:36:39.120657+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2563a639-3d63-4a4c-8034-77696b7b641c","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 32, \"completed_cases\": 468, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 468, \"run_group\": \"qwen3.5-4b-core-all\", \"run_id\": \"3a46714c-d579-4688-a8b0-558183b4e02a\", \"skipped_cases\": 0}","model_key":null,"run_id":"3a46714c-d579-4688-a8b0-558183b4e02a","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T16:36:39.120802+00:00"}]},"3b0292d4-8f8f-4843-900c-946c32476964":{"cells":[{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T15:45:29.838316+00:00","execution_signature":"sha256:execution:0caf2fff221781464037ba78ff0ed5a8158d632ab0e66481fc8df733dd8785c6","failure_count":0,"model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.5,"profile":"core","run_id":"3b0292d4-8f8f-4843-900c-946c32476964","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"3b0292d4-8f8f-4843-900c-946c32476964","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":2837072896,"deployment":"local","effective_model_id":"nemotron-nano-4b","execution_signature":"sha256:execution:0caf2fff221781464037ba78ff0ed5a8158d632ab0e66481fc8df733dd8785c6","family":"nemotron-nano","model_id":"nvidia/nemotron-3-nano-4b","model_key":"nemotron-nano-4b","model_signature":"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34","parameters_b":4.0,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"3b0292d4-8f8f-4843-900c-946c32476964","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:0caf2fff221781464037ba78ff0ed5a8158d632ab0e66481fc8df733dd8785c6","model_key":"nemotron-nano-4b","process_cpu_percent_avg":10.014690415304136,"process_rss_bytes_avg":4240616618.6666665,"process_rss_bytes_peak":4889542656.0,"sample_count":153,"sample_interval_ms":267.2538513759112,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5000953856.0}],"run":{"completed_at_utc":"2026-10-15T15:45:29.838316+00:00","created_at_utc":"2026-10-15T15:44:53.642093+00:00","git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","integrity_state":"VALID","manifest_checksum":"8689f0539fd58b45df0797e7a0dcba185b58377c77879dff339de517f2836559","profile":"core","run_group":"math-v3-core-nemotron-nano-4b","run_id":"3b0292d4-8f8f-4843-900c-946c32476964","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"3"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"64bcc7b7-f95e-4a7d-8337-397d8974164f","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"nemotron-nano-4b\"], \"profile\": \"core\", \"run_group\": \"math-v3-core-nemotron-nano-4b\", \"run_id\": \"3b0292d4-8f8f-4843-900c-946c32476964\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"3\"}","model_key":null,"run_id":"3b0292d4-8f8f-4843-900c-946c32476964","sample_id":null,"status":null,"timestamp_utc":"2026-10-15T15:44:53.642093+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"91f88024-e498-435f-bace-0dc8460ee7c7","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-core-nemotron-nano-4b\", \"run_id\": \"3b0292d4-8f8f-4843-900c-946c32476964\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"3b0292d4-8f8f-4843-900c-946c32476964","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T15:44:53.663969+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":3113.2328750099987,"error_message":null,"error_type":null,"event_id":"85b172cf-7cd4-4532-8b2e-7a1718dadc0d","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:0caf2fff221781464037ba78ff0ed5a8158d632ab0e66481fc8df733dd8785c6\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-core-nemotron-nano-4b\", \"run_id\": \"3b0292d4-8f8f-4843-900c-946c32476964\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 2837072896, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 42, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"nemotron-nano-4b\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"3b0292d4-8f8f-4843-900c-946c32476964","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T15:44:56.777812+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c051705f-b640-4d80-8c63-6db08157e596","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:0caf2fff221781464037ba78ff0ed5a8158d632ab0e66481fc8df733dd8785c6\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-core-nemotron-nano-4b\", \"run_id\": \"3b0292d4-8f8f-4843-900c-946c32476964\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 2837072896, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 42, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"nemotron-nano-4b\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"3b0292d4-8f8f-4843-900c-946c32476964","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T15:44:56.778428+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"70de7c9d-0062-43c9-82e8-47286e189f26","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:0caf2fff221781464037ba78ff0ed5a8158d632ab0e66481fc8df733dd8785c6\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-core-nemotron-nano-4b\", \"run_id\": \"3b0292d4-8f8f-4843-900c-946c32476964\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 2837072896, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 42, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"nemotron-nano-4b\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"3b0292d4-8f8f-4843-900c-946c32476964","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T15:44:56.804703+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"242a211d-1129-4609-ae3b-1ce88678ab8c","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:0caf2fff221781464037ba78ff0ed5a8158d632ab0e66481fc8df733dd8785c6\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-core-nemotron-nano-4b\", \"run_id\": \"3b0292d4-8f8f-4843-900c-946c32476964\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 2837072896, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 42, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"nemotron-nano-4b\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"3b0292d4-8f8f-4843-900c-946c32476964","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T15:45:29.486993+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"9cbf9175-cb75-400e-9b0a-95136f9981b3","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:0caf2fff221781464037ba78ff0ed5a8158d632ab0e66481fc8df733dd8785c6\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-core-nemotron-nano-4b\", \"run_id\": \"3b0292d4-8f8f-4843-900c-946c32476964\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 2837072896, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 42, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"nemotron-nano-4b\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"3b0292d4-8f8f-4843-900c-946c32476964","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T15:45:29.487180+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":347.0317500177771,"error_message":null,"error_type":null,"event_id":"bb1d0e75-b3d6-49d0-9f65-20c044038f8f","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:0caf2fff221781464037ba78ff0ed5a8158d632ab0e66481fc8df733dd8785c6\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-core-nemotron-nano-4b\", \"run_id\": \"3b0292d4-8f8f-4843-900c-946c32476964\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 2837072896, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 42, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"nemotron-nano-4b\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b","run_id":"3b0292d4-8f8f-4843-900c-946c32476964","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T15:45:29.834401+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a4c0be4a-2a3b-4626-85ea-c48be61e1251","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:0caf2fff221781464037ba78ff0ed5a8158d632ab0e66481fc8df733dd8785c6\", \"model_key\": \"nemotron-nano-4b\", \"model_signature\": \"sha256:model:54a82463d42fc07a6c3f5b296f1c61e569779f495d9fb5172f9feb8941bd2d34\", \"run_group\": \"math-v3-core-nemotron-nano-4b\", \"run_id\": \"3b0292d4-8f8f-4843-900c-946c32476964\"}","model_key":"nemotron-nano-4b","run_id":"3b0292d4-8f8f-4843-900c-946c32476964","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T15:45:29.838178+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ef86302b-dd28-4fcb-a257-92a16e2531f9","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 40, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 40, \"run_group\": \"math-v3-core-nemotron-nano-4b\", \"run_id\": \"3b0292d4-8f8f-4843-900c-946c32476964\", \"skipped_cases\": 0}","model_key":null,"run_id":"3b0292d4-8f8f-4843-900c-946c32476964","sample_id":null,"status":null,"timestamp_utc":"2026-10-15T15:45:29.838316+00:00"}]},"5623e5c1-213d-447a-88e9-ac51a69a92fa":{"cells":[{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T12:48:39.598492+00:00","execution_signature":"sha256:execution:ff14676ac39e23af360cd1209491ef1c14e3dab9a3198b48054de5c6de699fdb","failure_count":0,"model_key":"limite-1b-violetto","model_signature":"sha256:model:3fe1bfff5bb2c952f96bb799970eb76d9db17f9f880bc3326f0bdfbdd51577cc","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.0,"profile":"core","run_id":"5623e5c1-213d-447a-88e9-ac51a69a92fa","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"5623e5c1-213d-447a-88e9-ac51a69a92fa","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:ff14676ac39e23af360cd1209491ef1c14e3dab9a3198b48054de5c6de699fdb","family":"limite","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:3fe1bfff5bb2c952f96bb799970eb76d9db17f9f880bc3326f0bdfbdd51577cc","parameters_b":1.0,"provider_key":"vllm","quantization":null,"run_id":"5623e5c1-213d-447a-88e9-ac51a69a92fa","runtime_key":"vllm-local"}],"resources":[],"run":{"completed_at_utc":"2026-10-15T12:48:39.598492+00:00","created_at_utc":"2026-10-15T12:48:39.193018+00:00","git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","integrity_state":"VALID","manifest_checksum":"620d04e9f7cdb22a8cafd2cfd54d4ec4a7957ec63babf14de43fcd020b9eace4","profile":"core","run_group":"limite-mathematics","run_id":"5623e5c1-213d-447a-88e9-ac51a69a92fa","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"4a95b04d-ef3f-492f-862f-e5dcc7f3dc5e","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"limite-1b-violetto\"], \"profile\": \"core\", \"run_group\": \"limite-mathematics\", \"run_id\": \"5623e5c1-213d-447a-88e9-ac51a69a92fa\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"5623e5c1-213d-447a-88e9-ac51a69a92fa","sample_id":null,"status":null,"timestamp_utc":"2026-10-15T12:48:39.193018+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6900fdc6-6804-49ec-9f8f-c7ec9ffcf79a","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:3fe1bfff5bb2c952f96bb799970eb76d9db17f9f880bc3326f0bdfbdd51577cc\", \"provider_key\": \"vllm\", \"run_group\": \"limite-mathematics\", \"run_id\": \"5623e5c1-213d-447a-88e9-ac51a69a92fa\", \"runtime_key\": \"vllm-local\"}","model_key":"limite-1b-violetto","run_id":"5623e5c1-213d-447a-88e9-ac51a69a92fa","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T12:48:39.223182+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":53.61883295699954,"error_message":null,"error_type":null,"event_id":"f1977fd1-a0e3-40f0-ab9c-a7e2b8d28909","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:ff14676ac39e23af360cd1209491ef1c14e3dab9a3198b48054de5c6de699fdb\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:3fe1bfff5bb2c952f96bb799970eb76d9db17f9f880bc3326f0bdfbdd51577cc\", \"provider_key\": \"vllm\", \"run_group\": \"limite-mathematics\", \"run_id\": \"5623e5c1-213d-447a-88e9-ac51a69a92fa\", \"runtime_key\": \"vllm-local\"}","model_key":"limite-1b-violetto","run_id":"5623e5c1-213d-447a-88e9-ac51a69a92fa","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T12:48:39.277421+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"35cf261c-37f3-4025-a626-e2f5923c8e46","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:ff14676ac39e23af360cd1209491ef1c14e3dab9a3198b48054de5c6de699fdb\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:3fe1bfff5bb2c952f96bb799970eb76d9db17f9f880bc3326f0bdfbdd51577cc\", \"provider_key\": \"vllm\", \"run_group\": \"limite-mathematics\", \"run_id\": \"5623e5c1-213d-447a-88e9-ac51a69a92fa\", \"runtime_key\": \"vllm-local\"}","model_key":"limite-1b-violetto","run_id":"5623e5c1-213d-447a-88e9-ac51a69a92fa","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T12:48:39.277780+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a03a8fb2-10ce-4bd1-8bc1-e4b9dc88e7ab","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:ff14676ac39e23af360cd1209491ef1c14e3dab9a3198b48054de5c6de699fdb\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:3fe1bfff5bb2c952f96bb799970eb76d9db17f9f880bc3326f0bdfbdd51577cc\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"vllm\", \"run_group\": \"limite-mathematics\", \"run_id\": \"5623e5c1-213d-447a-88e9-ac51a69a92fa\", \"runtime_key\": \"vllm-local\"}","model_key":"limite-1b-violetto","run_id":"5623e5c1-213d-447a-88e9-ac51a69a92fa","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T12:48:39.294317+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"4c2bf872-1b8c-423c-b5ab-c8e9b8b2f34f","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:ff14676ac39e23af360cd1209491ef1c14e3dab9a3198b48054de5c6de699fdb\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:3fe1bfff5bb2c952f96bb799970eb76d9db17f9f880bc3326f0bdfbdd51577cc\", \"profile\": \"core\", \"provider_key\": \"vllm\", \"run_group\": \"limite-mathematics\", \"run_id\": \"5623e5c1-213d-447a-88e9-ac51a69a92fa\", \"runtime_key\": \"vllm-local\"}","model_key":"limite-1b-violetto","run_id":"5623e5c1-213d-447a-88e9-ac51a69a92fa","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T12:48:39.594185+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ce8dc9fc-6ad4-48e4-a400-8ec2da316fcf","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:ff14676ac39e23af360cd1209491ef1c14e3dab9a3198b48054de5c6de699fdb\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:3fe1bfff5bb2c952f96bb799970eb76d9db17f9f880bc3326f0bdfbdd51577cc\", \"provider_key\": \"vllm\", \"run_group\": \"limite-mathematics\", \"run_id\": \"5623e5c1-213d-447a-88e9-ac51a69a92fa\", \"runtime_key\": \"vllm-local\"}","model_key":"limite-1b-violetto","run_id":"5623e5c1-213d-447a-88e9-ac51a69a92fa","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T12:48:39.594514+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.006875023245811462,"error_message":null,"error_type":null,"event_id":"016665d9-4897-480f-a2fc-fd21915f1326","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:ff14676ac39e23af360cd1209491ef1c14e3dab9a3198b48054de5c6de699fdb\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:3fe1bfff5bb2c952f96bb799970eb76d9db17f9f880bc3326f0bdfbdd51577cc\", \"provider_key\": \"vllm\", \"run_group\": \"limite-mathematics\", \"run_id\": \"5623e5c1-213d-447a-88e9-ac51a69a92fa\", \"runtime_key\": \"vllm-local\"}","model_key":"limite-1b-violetto","run_id":"5623e5c1-213d-447a-88e9-ac51a69a92fa","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T12:48:39.594733+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"341483cf-cd68-4e87-a6c9-d354e64957a4","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:ff14676ac39e23af360cd1209491ef1c14e3dab9a3198b48054de5c6de699fdb\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:3fe1bfff5bb2c952f96bb799970eb76d9db17f9f880bc3326f0bdfbdd51577cc\", \"run_group\": \"limite-mathematics\", \"run_id\": \"5623e5c1-213d-447a-88e9-ac51a69a92fa\"}","model_key":"limite-1b-violetto","run_id":"5623e5c1-213d-447a-88e9-ac51a69a92fa","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T12:48:39.598234+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1434e36e-ea13-42f2-917c-3b6467e8b880","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 40, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 40, \"run_group\": \"limite-mathematics\", \"run_id\": \"5623e5c1-213d-447a-88e9-ac51a69a92fa\", \"skipped_cases\": 0}","model_key":null,"run_id":"5623e5c1-213d-447a-88e9-ac51a69a92fa","sample_id":null,"status":null,"timestamp_utc":"2026-10-15T12:48:39.598492+00:00"}]},"8eef4bab-617a-4638-9e51-53e72034e821":{"cells":[{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-17T21:24:56.752644+00:00","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","failure_count":0,"model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.8,"profile":"core","run_id":"8eef4bab-617a-4638-9e51-53e72034e821","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"8eef4bab-617a-4638-9e51-53e72034e821","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","family":"limite","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","parameters_b":1.0,"provider_key":"transformers","quantization":null,"run_id":"8eef4bab-617a-4638-9e51-53e72034e821","runtime_key":"transformers-local"}],"resources":[],"run":{"completed_at_utc":"2026-10-17T21:24:56.752644+00:00","created_at_utc":"2026-10-17T21:13:50.638554+00:00","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","integrity_state":"VALID","manifest_checksum":"6dfbf26de4ed844edc0871e5e69c778d861ee2115660f925c1a0fd3c7dd9b5d9","profile":"core","run_group":"limite-violetto-v2-core-2048","run_id":"8eef4bab-617a-4638-9e51-53e72034e821","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"3"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e6825f24-598b-4cce-94e7-39221d12a66c","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning-v2\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"limite-1b-violetto\"], \"profile\": \"core\", \"run_group\": \"limite-violetto-v2-core-2048\", \"run_id\": \"8eef4bab-617a-4638-9e51-53e72034e821\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"3\"}","model_key":null,"run_id":"8eef4bab-617a-4638-9e51-53e72034e821","sample_id":null,"status":null,"timestamp_utc":"2026-10-17T21:13:50.638554+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"27e7cebc-48f0-4325-a57e-3bb499589652","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"provider_key\": \"transformers\", \"run_group\": \"limite-violetto-v2-core-2048\", \"run_id\": \"8eef4bab-617a-4638-9e51-53e72034e821\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"8eef4bab-617a-4638-9e51-53e72034e821","sample_id":null,"status":"started","timestamp_utc":"2026-10-17T21:13:50.667476+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.3206670517101884,"error_message":null,"error_type":null,"event_id":"d2668fc3-fb86-47cd-95c4-7890dc6137b0","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"provider_key\": \"transformers\", \"run_group\": \"limite-violetto-v2-core-2048\", \"run_id\": \"8eef4bab-617a-4638-9e51-53e72034e821\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"8eef4bab-617a-4638-9e51-53e72034e821","sample_id":null,"status":"completed","timestamp_utc":"2026-10-17T21:13:50.668463+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"cad480e3-484e-40c2-8723-4ea18c0617da","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning-v2\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"provider_key\": \"transformers\", \"run_group\": \"limite-violetto-v2-core-2048\", \"run_id\": \"8eef4bab-617a-4638-9e51-53e72034e821\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"8eef4bab-617a-4638-9e51-53e72034e821","sample_id":null,"status":"started","timestamp_utc":"2026-10-17T21:13:50.668712+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ff97aa59-cac8-4fe3-8458-a6332b5b586a","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:738d15e8d280a7feb3ee7f9fda46cfc9b6c4515af2236f69dde793dbdbe30496\", \"capability_id\": \"mathematical-reasoning-v2\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"transformers\", \"run_group\": \"limite-violetto-v2-core-2048\", \"run_id\": \"8eef4bab-617a-4638-9e51-53e72034e821\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"8eef4bab-617a-4638-9e51-53e72034e821","sample_id":null,"status":"started","timestamp_utc":"2026-10-17T21:13:50.682786+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"bde7759b-0c41-4836-99df-94c9dce06f3b","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:738d15e8d280a7feb3ee7f9fda46cfc9b6c4515af2236f69dde793dbdbe30496\", \"capability_id\": \"mathematical-reasoning-v2\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"profile\": \"core\", \"provider_key\": \"transformers\", \"run_group\": \"limite-violetto-v2-core-2048\", \"run_id\": \"8eef4bab-617a-4638-9e51-53e72034e821\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"8eef4bab-617a-4638-9e51-53e72034e821","sample_id":null,"status":"completed","timestamp_utc":"2026-10-17T21:24:56.391996+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"19ffa260-b2f4-490e-a022-6bf983a2a263","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"provider_key\": \"transformers\", \"run_group\": \"limite-violetto-v2-core-2048\", \"run_id\": \"8eef4bab-617a-4638-9e51-53e72034e821\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"8eef4bab-617a-4638-9e51-53e72034e821","sample_id":null,"status":"started","timestamp_utc":"2026-10-17T21:24:56.392253+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":353.7606250029057,"error_message":null,"error_type":null,"event_id":"f45b6be6-381e-4584-94a5-9e13333321f8","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"provider_key\": \"transformers\", \"run_group\": \"limite-violetto-v2-core-2048\", \"run_id\": \"8eef4bab-617a-4638-9e51-53e72034e821\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"8eef4bab-617a-4638-9e51-53e72034e821","sample_id":null,"status":"completed","timestamp_utc":"2026-10-17T21:24:56.746258+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f2d1c9c4-36fd-42ff-bc47-c67cf77c393d","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:738d15e8d280a7feb3ee7f9fda46cfc9b6c4515af2236f69dde793dbdbe30496\", \"capability_id\": \"mathematical-reasoning-v2\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"run_group\": \"limite-violetto-v2-core-2048\", \"run_id\": \"8eef4bab-617a-4638-9e51-53e72034e821\"}","model_key":"limite-1b-violetto","run_id":"8eef4bab-617a-4638-9e51-53e72034e821","sample_id":null,"status":"completed","timestamp_utc":"2026-10-17T21:24:56.752110+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"272fbb58-cc5b-4124-bfee-8fde3d6a96b2","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 40, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 40, \"run_group\": \"limite-violetto-v2-core-2048\", \"run_id\": \"8eef4bab-617a-4638-9e51-53e72034e821\", \"skipped_cases\": 0}","model_key":null,"run_id":"8eef4bab-617a-4638-9e51-53e72034e821","sample_id":null,"status":null,"timestamp_utc":"2026-10-17T21:24:56.752644+00:00"}]},"954053b1-6f87-4196-bb2a-028404b6fef1":{"cells":[{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T13:46:02.371339+00:00","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","failure_count":0,"model_key":"limite-1b-violetto","model_signature":"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.725,"profile":"core","run_id":"954053b1-6f87-4196-bb2a-028404b6fef1","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"954053b1-6f87-4196-bb2a-028404b6fef1","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","family":"limite","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65","parameters_b":1.0,"provider_key":"transformers","quantization":null,"run_id":"954053b1-6f87-4196-bb2a-028404b6fef1","runtime_key":"transformers-local"}],"resources":[],"run":{"completed_at_utc":"2026-10-15T13:46:02.371339+00:00","created_at_utc":"2026-10-15T13:39:01.635060+00:00","git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","integrity_state":"VALID","manifest_checksum":"a6878e06dd0b7935fe0de6edb8f625b2408aaa834b0cbab2d6036ef84e23d4ac","profile":"core","run_group":"limite-violetto-core","run_id":"954053b1-6f87-4196-bb2a-028404b6fef1","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1f3d283c-b266-4eeb-9adf-5cae5898c6b5","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"limite-1b-violetto\"], \"profile\": \"core\", \"run_group\": \"limite-violetto-core\", \"run_id\": \"954053b1-6f87-4196-bb2a-028404b6fef1\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"954053b1-6f87-4196-bb2a-028404b6fef1","sample_id":null,"status":null,"timestamp_utc":"2026-10-15T13:39:01.635060+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6ea3069c-5846-4f14-af90-737bf7ce2ccd","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65\", \"provider_key\": \"transformers\", \"run_group\": \"limite-violetto-core\", \"run_id\": \"954053b1-6f87-4196-bb2a-028404b6fef1\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"954053b1-6f87-4196-bb2a-028404b6fef1","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T13:39:01.656125+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.2802090020850301,"error_message":null,"error_type":null,"event_id":"9cdb4570-7216-4bfc-a793-7d1507f0577f","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65\", \"provider_key\": \"transformers\", \"run_group\": \"limite-violetto-core\", \"run_id\": \"954053b1-6f87-4196-bb2a-028404b6fef1\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"954053b1-6f87-4196-bb2a-028404b6fef1","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T13:39:01.656967+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8f50cfab-04ca-455e-94e3-e4ec7a7a0749","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65\", \"provider_key\": \"transformers\", \"run_group\": \"limite-violetto-core\", \"run_id\": \"954053b1-6f87-4196-bb2a-028404b6fef1\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"954053b1-6f87-4196-bb2a-028404b6fef1","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T13:39:01.657229+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f5814b00-7791-4355-b468-89d5a58bc9dc","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"transformers\", \"run_group\": \"limite-violetto-core\", \"run_id\": \"954053b1-6f87-4196-bb2a-028404b6fef1\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"954053b1-6f87-4196-bb2a-028404b6fef1","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T13:39:01.671078+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"cb6ca740-aee7-4bb9-91a5-b5d2e0d0eb36","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65\", \"profile\": \"core\", \"provider_key\": \"transformers\", \"run_group\": \"limite-violetto-core\", \"run_id\": \"954053b1-6f87-4196-bb2a-028404b6fef1\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"954053b1-6f87-4196-bb2a-028404b6fef1","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T13:46:02.086415+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"54a36975-337f-440c-946d-3374a81d1385","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65\", \"provider_key\": \"transformers\", \"run_group\": \"limite-violetto-core\", \"run_id\": \"954053b1-6f87-4196-bb2a-028404b6fef1\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"954053b1-6f87-4196-bb2a-028404b6fef1","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T13:46:02.086688+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":277.8879589168355,"error_message":null,"error_type":null,"event_id":"1d8086aa-235d-438e-badb-e5157de3212f","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65\", \"provider_key\": \"transformers\", \"run_group\": \"limite-violetto-core\", \"run_id\": \"954053b1-6f87-4196-bb2a-028404b6fef1\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"954053b1-6f87-4196-bb2a-028404b6fef1","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T13:46:02.364886+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7d9239ad-6431-4459-ab13-abc9d0d8cf0d","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65\", \"run_group\": \"limite-violetto-core\", \"run_id\": \"954053b1-6f87-4196-bb2a-028404b6fef1\"}","model_key":"limite-1b-violetto","run_id":"954053b1-6f87-4196-bb2a-028404b6fef1","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T13:46:02.371139+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"772aedf5-374e-43aa-bd99-2eba3ca2f929","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 40, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 40, \"run_group\": \"limite-violetto-core\", \"run_id\": \"954053b1-6f87-4196-bb2a-028404b6fef1\", \"skipped_cases\": 0}","model_key":null,"run_id":"954053b1-6f87-4196-bb2a-028404b6fef1","sample_id":null,"status":null,"timestamp_utc":"2026-10-15T13:46:02.371339+00:00"}]},"9fb9f932-1f29-4646-b337-f33476d841ba":{"cells":[{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T15:44:53.050810+00:00","execution_signature":"sha256:execution:8190c8f3bb1f7b51ef597577c913ceae05f8cb7998a7d5221e848f67071166c9","failure_count":0,"model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.25,"profile":"core","run_id":"9fb9f932-1f29-4646-b337-f33476d841ba","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"9fb9f932-1f29-4646-b337-f33476d841ba","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":1280835840,"deployment":"local","effective_model_id":"qwen3.5-2b-q4km","execution_signature":"sha256:execution:8190c8f3bb1f7b51ef597577c913ceae05f8cb7998a7d5221e848f67071166c9","family":"qwen3.5","model_id":"unsloth/qwen3.5-2b","model_key":"qwen3.5-2b-q4km","model_signature":"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694","parameters_b":2.0,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"9fb9f932-1f29-4646-b337-f33476d841ba","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:8190c8f3bb1f7b51ef597577c913ceae05f8cb7998a7d5221e848f67071166c9","model_key":"qwen3.5-2b-q4km","process_cpu_percent_avg":8.953256029208966,"process_rss_bytes_avg":2160430830.9333334,"process_rss_bytes_peak":2483290112.0,"sample_count":94,"sample_interval_ms":267.4467081844341,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5225758720.0}],"run":{"completed_at_utc":"2026-10-15T15:44:53.050810+00:00","created_at_utc":"2026-10-15T15:44:34.529590+00:00","git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","integrity_state":"VALID","manifest_checksum":"bbf8a345656e3ceb4caed976e33747a28f1fe2c2d4359acbe28bc5722beaed48","profile":"core","run_group":"math-v3-core-qwen3.5-2b-q4km","run_id":"9fb9f932-1f29-4646-b337-f33476d841ba","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"3"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6b837172-a436-4849-bd24-e998fe0c45ad","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"qwen3.5-2b-q4km\"], \"profile\": \"core\", \"run_group\": \"math-v3-core-qwen3.5-2b-q4km\", \"run_id\": \"9fb9f932-1f29-4646-b337-f33476d841ba\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"3\"}","model_key":null,"run_id":"9fb9f932-1f29-4646-b337-f33476d841ba","sample_id":null,"status":null,"timestamp_utc":"2026-10-15T15:44:34.529590+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ad30d327-d861-4a10-9f25-ece3e25b080b","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-core-qwen3.5-2b-q4km\", \"run_id\": \"9fb9f932-1f29-4646-b337-f33476d841ba\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"9fb9f932-1f29-4646-b337-f33476d841ba","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T15:44:34.556581+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2116.0324160009623,"error_message":null,"error_type":null,"event_id":"d5c96314-98a3-4cd3-87c5-63c05da1479a","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:8190c8f3bb1f7b51ef597577c913ceae05f8cb7998a7d5221e848f67071166c9\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-core-qwen3.5-2b-q4km\", \"run_id\": \"9fb9f932-1f29-4646-b337-f33476d841ba\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 1280835840, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"9fb9f932-1f29-4646-b337-f33476d841ba","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T15:44:36.673775+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"162f58e3-be40-4e61-9d70-76119ebb1dfe","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:8190c8f3bb1f7b51ef597577c913ceae05f8cb7998a7d5221e848f67071166c9\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-core-qwen3.5-2b-q4km\", \"run_id\": \"9fb9f932-1f29-4646-b337-f33476d841ba\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 1280835840, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"9fb9f932-1f29-4646-b337-f33476d841ba","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T15:44:36.673987+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"5c77985f-60ec-4f94-9714-4d6f084b3d29","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:8190c8f3bb1f7b51ef597577c913ceae05f8cb7998a7d5221e848f67071166c9\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-core-qwen3.5-2b-q4km\", \"run_id\": \"9fb9f932-1f29-4646-b337-f33476d841ba\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 1280835840, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"9fb9f932-1f29-4646-b337-f33476d841ba","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T15:44:36.690908+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2a97de1a-4efe-41aa-b1e7-90826f660ca7","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:8190c8f3bb1f7b51ef597577c913ceae05f8cb7998a7d5221e848f67071166c9\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-core-qwen3.5-2b-q4km\", \"run_id\": \"9fb9f932-1f29-4646-b337-f33476d841ba\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 1280835840, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"9fb9f932-1f29-4646-b337-f33476d841ba","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T15:44:52.912838+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f404dd46-5d7a-4bb8-9cdb-8294f5f42c0b","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:8190c8f3bb1f7b51ef597577c913ceae05f8cb7998a7d5221e848f67071166c9\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-core-qwen3.5-2b-q4km\", \"run_id\": \"9fb9f932-1f29-4646-b337-f33476d841ba\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 1280835840, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"9fb9f932-1f29-4646-b337-f33476d841ba","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T15:44:52.913040+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":133.31941608339548,"error_message":null,"error_type":null,"event_id":"c8579ceb-4592-417b-9eae-6cd0e80a93c6","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:8190c8f3bb1f7b51ef597577c913ceae05f8cb7998a7d5221e848f67071166c9\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-core-qwen3.5-2b-q4km\", \"run_id\": \"9fb9f932-1f29-4646-b337-f33476d841ba\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 1280835840, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-q4km","run_id":"9fb9f932-1f29-4646-b337-f33476d841ba","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T15:44:53.046567+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"31f00b23-5415-45b2-b6c7-52f2064a6ba6","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:8190c8f3bb1f7b51ef597577c913ceae05f8cb7998a7d5221e848f67071166c9\", \"model_key\": \"qwen3.5-2b-q4km\", \"model_signature\": \"sha256:model:efccb81c5afd40698fa318d2e5307d13b9951c7c77c83f27b5bc5c125c408694\", \"run_group\": \"math-v3-core-qwen3.5-2b-q4km\", \"run_id\": \"9fb9f932-1f29-4646-b337-f33476d841ba\"}","model_key":"qwen3.5-2b-q4km","run_id":"9fb9f932-1f29-4646-b337-f33476d841ba","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T15:44:53.050662+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"28b8eee6-4eac-4e83-8d76-7ecd7ea36713","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 40, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 40, \"run_group\": \"math-v3-core-qwen3.5-2b-q4km\", \"run_id\": \"9fb9f932-1f29-4646-b337-f33476d841ba\", \"skipped_cases\": 0}","model_key":null,"run_id":"9fb9f932-1f29-4646-b337-f33476d841ba","sample_id":null,"status":null,"timestamp_utc":"2026-10-15T15:44:53.050810+00:00"}]},"a97505cf-abc6-4782-bf99-38de2a4492b7":{"cells":[{"benchmark_signature":"sha256:benchmark:718e761e4ba3410c289a5bba91a6474827caaa8da30e644b3df925ba4d16fbfc","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T15:11:29.046506+00:00","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","failure_count":0,"model_key":"decisio-qwen3.5-4b-q4km","model_signature":"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6753246753246753,"profile":"smoke","run_id":"a97505cf-abc6-4782-bf99-38de2a4492b7","sample_count":77,"status":"COMPLETED","task_id":"intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"a97505cf-abc6-4782-bf99-38de2a4492b7","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","family":null,"model_id":"unsloth/qwen3.5-4b","model_key":"decisio-qwen3.5-4b-q4km","model_signature":"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a","parameters_b":null,"provider_key":"decisio","quantization":"Q4_K_M","run_id":"a97505cf-abc6-4782-bf99-38de2a4492b7","runtime_key":"decisio-local"}],"resources":[],"run":{"completed_at_utc":"2026-10-05T15:11:29.046506+00:00","created_at_utc":"2026-10-05T15:07:45.730296+00:00","git_commit":"2e7b64ca38a2db66a9d8b36910d2b2ac5da0c7ce","integrity_state":"VALID","manifest_checksum":"b109844e3071057d96a5ed478021475b1bdcfcb55d4d01fcaef127aa877bcb6a","profile":"smoke","run_group":"decisio-classification-20261005_170744","run_id":"a97505cf-abc6-4782-bf99-38de2a4492b7","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2d6f4594-939c-4627-91cf-6eb19e39ce55","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"intent-classification\"], \"model_keys\": [\"decisio-qwen3.5-4b-q4km\"], \"profile\": \"smoke\", \"run_group\": \"decisio-classification-20261005_170744\", \"run_id\": \"a97505cf-abc6-4782-bf99-38de2a4492b7\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"a97505cf-abc6-4782-bf99-38de2a4492b7","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T15:07:45.730296+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"34d6c797-96e6-43f7-8ab7-7f90770b56d5","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_170744\", \"run_id\": \"a97505cf-abc6-4782-bf99-38de2a4492b7\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-4b-q4km","run_id":"a97505cf-abc6-4782-bf99-38de2a4492b7","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T15:07:45.750455+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":4055.564541995409,"error_message":null,"error_type":null,"event_id":"a9803fa0-eb19-4bb2-aa4b-c3aa61094cdd","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_170744\", \"run_id\": \"a97505cf-abc6-4782-bf99-38de2a4492b7\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-4b-q4km","run_id":"a97505cf-abc6-4782-bf99-38de2a4492b7","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T15:07:49.806502+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b65009c8-b0ec-4e21-9dcb-5a829f00cff9","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_170744\", \"run_id\": \"a97505cf-abc6-4782-bf99-38de2a4492b7\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-4b-q4km","run_id":"a97505cf-abc6-4782-bf99-38de2a4492b7","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T15:07:49.807099+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"00cb8271-6750-4998-aa1a-4bf6128cf454","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:718e761e4ba3410c289a5bba91a6474827caaa8da30e644b3df925ba4d16fbfc\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a\", \"planned_cases\": 77, \"profile\": \"smoke\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_170744\", \"run_id\": \"a97505cf-abc6-4782-bf99-38de2a4492b7\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-4b-q4km","run_id":"a97505cf-abc6-4782-bf99-38de2a4492b7","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T15:07:49.812316+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8cf386b1-2a89-4ed7-8fb3-735981038072","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:718e761e4ba3410c289a5bba91a6474827caaa8da30e644b3df925ba4d16fbfc\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a\", \"profile\": \"smoke\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_170744\", \"run_id\": \"a97505cf-abc6-4782-bf99-38de2a4492b7\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-4b-q4km","run_id":"a97505cf-abc6-4782-bf99-38de2a4492b7","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T15:11:29.040556+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f5ffcf4b-e1bb-4f69-ab9d-ba1d5c28590e","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_170744\", \"run_id\": \"a97505cf-abc6-4782-bf99-38de2a4492b7\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-4b-q4km","run_id":"a97505cf-abc6-4782-bf99-38de2a4492b7","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T15:11:29.040726+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.011166994227096438,"error_message":null,"error_type":null,"event_id":"b0ca2a0d-da1d-4430-87cd-1b4746436765","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_170744\", \"run_id\": \"a97505cf-abc6-4782-bf99-38de2a4492b7\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-4b-q4km","run_id":"a97505cf-abc6-4782-bf99-38de2a4492b7","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T15:11:29.040922+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"5ab8383c-f38f-4f0c-ad11-ebf1816e8068","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:718e761e4ba3410c289a5bba91a6474827caaa8da30e644b3df925ba4d16fbfc\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a\", \"run_group\": \"decisio-classification-20261005_170744\", \"run_id\": \"a97505cf-abc6-4782-bf99-38de2a4492b7\"}","model_key":"decisio-qwen3.5-4b-q4km","run_id":"a97505cf-abc6-4782-bf99-38de2a4492b7","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T15:11:29.046325+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d6974fd5-624f-4b13-8514-fd5a4b5cfd2f","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 77, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 77, \"run_group\": \"decisio-classification-20261005_170744\", \"run_id\": \"a97505cf-abc6-4782-bf99-38de2a4492b7\", \"skipped_cases\": 0}","model_key":null,"run_id":"a97505cf-abc6-4782-bf99-38de2a4492b7","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T15:11:29.046506+00:00"}]},"b10ad834-2bdd-409b-9073-49f0233792ea":{"cells":[{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","execution_signature":"sha256:execution:80688f79e2c27746cabb9c8e268a98d63ee3d732cb5c73d70fc581a1958017fc","failure_count":0,"model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.475,"profile":"core","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":0,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.125,"profile":"core","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","execution_signature":"sha256:execution:89a74e35df26090fe2c6adb16b01e32373fbcdad6966a1b23cc9b88ba35a5170","failure_count":0,"model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.475,"profile":"core","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","execution_signature":"sha256:execution:c58f103ecf1090aab4510c6708cd9446fefc4137c61ae9ca33da8cc1c5d3faf8","failure_count":0,"model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.1,"profile":"core","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","failure_count":0,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.25,"profile":"core","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","execution_signature":"sha256:execution:d6ef7d6b6d33f70d69c5d1df4568941686a610cda0bde994801472a902500dd4","failure_count":0,"model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.4,"profile":"core","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","failure_count":0,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.475,"profile":"core","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","execution_signature":"sha256:execution:9d728f687c9bb0f47a2c5a8c0e61b6da5a7e2833ad1afcffb18c7b4eb2767886","failure_count":0,"model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.5,"profile":"core","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","execution_signature":"sha256:execution:1b1fd8e53d6820b3449bce86d39dff358b1d6b5726cab7289358ad7888484aa8","failure_count":0,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":null,"profile":"core","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":0,"status":"PARTIAL","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:53:37.155498+00:00","execution_signature":"sha256:execution:1b1fd8e53d6820b3449bce86d39dff358b1d6b5726cab7289358ad7888484aa8","failure_count":0,"model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":null,"profile":"core","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_count":0,"status":"PARTIAL","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_json":"{}","sweep_id":null},"failure_summary":[{"count":2,"error_type":"TransportError","event_type":"model.prepare.failed","first_at":"2026-10-16T08:51:58.119416+00:00","last_at":"2026-10-16T08:53:37.117384+00:00"}],"models":[{"artifact_format":"gguf","artifact_size_bytes":3803452480,"deployment":"local","effective_model_id":"bonsai-27b-q10","execution_signature":"sha256:execution:80688f79e2c27746cabb9c8e268a98d63ee3d732cb5c73d70fc581a1958017fc","family":"bonsai","model_id":"prism-ml/bonsai-27b","model_key":"bonsai-27b-q10","model_signature":"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d","parameters_b":27.0,"provider_key":"korgis","quantization":"Q1_0","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local"},{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"gemini-3.5-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","family":"gemini-3.5","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","parameters_b":null,"provider_key":"gemini","quantization":null,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"gemini-api"},{"artifact_format":"gguf","artifact_size_bytes":4233677952,"deployment":"local","effective_model_id":"nemotron-nano-4b-q8","execution_signature":"sha256:execution:89a74e35df26090fe2c6adb16b01e32373fbcdad6966a1b23cc9b88ba35a5170","family":"nemotron-nano","model_id":"nvidia/nemotron-3-nano-4b-q8","model_key":"nemotron-nano-4b-q8","model_signature":"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e","parameters_b":4.0,"provider_key":"korgis","quantization":"Q8_0","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local"},{"artifact_format":"gguf","artifact_size_bytes":532517120,"deployment":"local","effective_model_id":"qwen3.5-0.8b-q4km","execution_signature":"sha256:execution:c58f103ecf1090aab4510c6708cd9446fefc4137c61ae9ca33da8cc1c5d3faf8","family":"qwen3.5","model_id":"unsloth/qwen3.5-0.8b","model_key":"qwen3.5-0.8b-q4km","model_signature":"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866","parameters_b":0.8,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local"},{"artifact_format":"gguf","artifact_size_bytes":3775709216,"deployment":"local","effective_model_id":"qwen3.5-2b-bf16","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","family":"qwen3.5","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","parameters_b":2.0,"provider_key":"korgis","quantization":"BF16","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local"},{"artifact_format":"gguf","artifact_size_bytes":2740937888,"deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:d6ef7d6b6d33f70d69c5d1df4568941686a610cda0bde994801472a902500dd4","family":"qwen3.5","model_id":"unsloth/qwen3.5-4b","model_key":"qwen3.5-4b-q4km","model_signature":"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481","parameters_b":4.0,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local"},{"artifact_format":"gguf","artifact_size_bytes":4482403488,"deployment":"local","effective_model_id":"qwen3.5-4b-q8","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","family":"qwen3.5","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","parameters_b":4.0,"provider_key":"korgis","quantization":"Q8_0","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local"},{"artifact_format":"gguf","artifact_size_bytes":5627044256,"deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:9d728f687c9bb0f47a2c5a8c0e61b6da5a7e2833ad1afcffb18c7b4eb2767886","family":"qwen3.5","model_id":"qwen/qwen3.5-9b","model_key":"qwen3.5-9b-q4km","model_signature":"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42","parameters_b":9.0,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local"},{"artifact_format":"gguf","artifact_size_bytes":4121781472,"deployment":"local","effective_model_id":"qwen3.5-9b-ud-q2kxl","execution_signature":"sha256:execution:1b1fd8e53d6820b3449bce86d39dff358b1d6b5726cab7289358ad7888484aa8","family":"qwen3.5","model_id":"qwen/qwen3.5-9b-ud-q2kxl","model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","parameters_b":9.0,"provider_key":"korgis","quantization":"UD-Q2_K_XL","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local"},{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.8-27b-iq1s","execution_signature":"sha256:execution:1b1fd8e53d6820b3449bce86d39dff358b1d6b5726cab7289358ad7888484aa8","family":"qwen3.8","model_id":"Qwen/Qwen3.8-27B","model_key":"qwen3.8-27b-iq1s","model_signature":"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043","parameters_b":27.0,"provider_key":"korgis","quantization":"UD-IQ1_S","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:80688f79e2c27746cabb9c8e268a98d63ee3d732cb5c73d70fc581a1958017fc","model_key":"bonsai-27b-q10","process_cpu_percent_avg":10.004219049002119,"process_rss_bytes_avg":7178716753.609695,"process_rss_bytes_peak":9109520384.0,"sample_count":342,"sample_interval_ms":299.5298028563516,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5797576704.0},{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:89a74e35df26090fe2c6adb16b01e32373fbcdad6966a1b23cc9b88ba35a5170","model_key":"nemotron-nano-4b-q8","process_cpu_percent_avg":8.818030339465103,"process_rss_bytes_avg":7560277067.093333,"process_rss_bytes_peak":10620682240.0,"sample_count":152,"sample_interval_ms":260.26064477991895,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":10035691520.0},{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:c58f103ecf1090aab4510c6708cd9446fefc4137c61ae9ca33da8cc1c5d3faf8","model_key":"qwen3.5-0.8b-q4km","process_cpu_percent_avg":15.698688370479838,"process_rss_bytes_avg":1544970035.2,"process_rss_bytes_peak":2028224512.0,"sample_count":80,"sample_interval_ms":263.3856863947585,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5715886080.0},{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","model_key":"qwen3.5-2b-bf16","process_cpu_percent_avg":8.718440372188164,"process_rss_bytes_avg":4446060373.333334,"process_rss_bytes_peak":4834885632.0,"sample_count":121,"sample_interval_ms":268.43950018325506,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5017370624.0},{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:d6ef7d6b6d33f70d69c5d1df4568941686a610cda0bde994801472a902500dd4","model_key":"qwen3.5-4b-q4km","process_cpu_percent_avg":10.67374362291546,"process_rss_bytes_avg":4568710143.999999,"process_rss_bytes_peak":5363400704.0,"sample_count":129,"sample_interval_ms":303.8685138652605,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5199380480.0},{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","model_key":"qwen3.5-4b-q8","process_cpu_percent_avg":10.86561479591514,"process_rss_bytes_avg":5559348804.266666,"process_rss_bytes_peak":6319587328.0,"sample_count":146,"sample_interval_ms":295.3429861280407,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4841750528.0},{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:9d728f687c9bb0f47a2c5a8c0e61b6da5a7e2833ad1afcffb18c7b4eb2767886","model_key":"qwen3.5-9b-q4km","process_cpu_percent_avg":7.04412305796482,"process_rss_bytes_avg":8143566028.8,"process_rss_bytes_peak":9486303232.0,"sample_count":176,"sample_interval_ms":291.7408815034529,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":9481781248.0}],"run":{"completed_at_utc":"2026-10-16T08:53:37.155498+00:00","created_at_utc":"2026-10-16T08:48:21.547929+00:00","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","integrity_state":"VALID","manifest_checksum":"cbf655fe3913c4b4acf7088ce35e251c6ed16469a8eb0b2d910f044ee52964e8","profile":"core","run_group":"math-v3-complete","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","seed":42,"source_dir":null,"status":"PARTIAL","suite_id":"capability-core","suite_version":"3"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e4827d70-842b-4a97-904f-6312d2efef2d","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"gemini-3.5-flash\", \"qwen3.5-0.8b-q4km\", \"qwen3.5-2b-bf16\", \"qwen3.5-4b-q4km\", \"qwen3.5-4b-q8\", \"nemotron-nano-4b-q8\", \"qwen3.5-9b-q4km\", \"qwen3.5-9b-ud-q2kxl\", \"bonsai-27b-q10\", \"qwen3.8-27b-iq1s\"], \"profile\": \"core\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"3\"}","model_key":null,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":null,"timestamp_utc":"2026-10-16T08:48:21.547929+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"dfe97d74-1964-43a2-b932-9cae4931c51b","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:48:21.577788+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":56.674291961826384,"error_message":null,"error_type":null,"event_id":"eb9bac3d-7db3-4894-8d30-ca89b8801803","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:48:21.635098+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a18737fa-ce7c-46ee-a4ee-9309ada43f5c","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:48:21.635473+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b9930c6e-cf5e-4170-9b98-3f1cf71749b2","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"gemini\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:48:21.662354+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"47b58fb9-d232-4d09-a54f-94613c3f8e12","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"profile\": \"core\", \"provider_key\": \"gemini\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:48:39.358833+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"710f05f2-f18e-417d-a45d-3897672630c9","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:48:39.362495+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.04045804962515831,"error_message":null,"error_type":null,"event_id":"8e6f7304-26ef-4cec-b93d-2a57ea4c2e86","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:48:39.363942+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"738844d4-4f6b-499a-96a9-5039ab36417f","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:48:39.366775+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2075.6182079203427,"error_message":null,"error_type":null,"event_id":"becb214c-ff10-4475-9156-c031917408a6","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:c58f103ecf1090aab4510c6708cd9446fefc4137c61ae9ca33da8cc1c5d3faf8\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 532517120, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-0.8b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:48:41.443364+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1ddbece0-c98c-4c1a-b08d-ae35785c14bb","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:c58f103ecf1090aab4510c6708cd9446fefc4137c61ae9ca33da8cc1c5d3faf8\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 532517120, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-0.8b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:48:41.444106+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"96a9f719-d62e-49a7-8756-820ea7ab4805","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:c58f103ecf1090aab4510c6708cd9446fefc4137c61ae9ca33da8cc1c5d3faf8\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 532517120, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-0.8b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:48:41.444693+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7edc9448-3876-4069-857e-937ee45e0acb","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:c58f103ecf1090aab4510c6708cd9446fefc4137c61ae9ca33da8cc1c5d3faf8\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 532517120, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-0.8b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:48:55.478044+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"39ab976d-6e7e-40b5-af9f-a74729106af4","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:c58f103ecf1090aab4510c6708cd9446fefc4137c61ae9ca33da8cc1c5d3faf8\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 532517120, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-0.8b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:48:55.478317+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":184.62149996776134,"error_message":null,"error_type":null,"event_id":"8d801fc8-7a00-4485-9f2d-2ea661d230bc","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:c58f103ecf1090aab4510c6708cd9446fefc4137c61ae9ca33da8cc1c5d3faf8\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 532517120, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-0.8b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:48:55.663232+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ffac4d44-ac1f-477f-a4aa-5dff3b9a2d8c","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:48:55.663949+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":4067.486249958165,"error_message":null,"error_type":null,"event_id":"22ce7812-e009-4954-b57a-1b091b4e4b9c","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:48:59.731853+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c7c2c04f-5df1-42e0-8dd8-3d60f3b25b6c","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:48:59.733069+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8b900125-9d72-49f9-818d-d3be92d0151b","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:48:59.733847+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f2a0a933-5f75-4826-93f2-5f0b437d7bf7","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:49:25.351836+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"10543d1c-7dec-4a91-afca-0f9d99241fcb","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:49:25.352358+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":401.5152909560129,"error_message":null,"error_type":null,"event_id":"69563c3d-d6c0-41ba-9273-cee56f1e19db","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:49:25.754239+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"fbbafa8a-0879-47fb-ac9e-18027afdcafc","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q4km","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:49:25.755391+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":3072.1053750021383,"error_message":null,"error_type":null,"event_id":"ee4e1fd6-aefe-42d2-95c5-af1673ff1d73","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:d6ef7d6b6d33f70d69c5d1df4568941686a610cda0bde994801472a902500dd4\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 2740937888, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_identity\": {\"captured_at\": 1792140568.820449, \"fingerprint\": \"6f6d804e077b349f6a17736fedffb2759453ac4ce364dca28c7ac8b002277972\", \"identity\": {\"artifact_key\": \"6db6031159a6d3e9f72a7c5b501900908f1c70737d61a02c5c6c7eb65da853f8\", \"backend\": {\"implementation\": \"StreamContractEngine\", \"name\": \"llama_server\", \"version\": \"build-10964@b29c606e2\"}, \"config_digest\": \"2874b30db323775541789cb8e3759427078d849263afe91839813790744df509\", \"hardware_key\": \"40a8c1371e06ac920aa3835029c4b3825f48de6152fdb15ad9c18f572a80b356\", \"schema_version\": 1}}, \"runtime_key\": \"qwen3.5-4b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q4km","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:49:28.827919+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"81c14b8f-890c-46ce-a446-9102f3dff37b","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:d6ef7d6b6d33f70d69c5d1df4568941686a610cda0bde994801472a902500dd4\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 2740937888, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_identity\": {\"captured_at\": 1792140568.820449, \"fingerprint\": \"6f6d804e077b349f6a17736fedffb2759453ac4ce364dca28c7ac8b002277972\", \"identity\": {\"artifact_key\": \"6db6031159a6d3e9f72a7c5b501900908f1c70737d61a02c5c6c7eb65da853f8\", \"backend\": {\"implementation\": \"StreamContractEngine\", \"name\": \"llama_server\", \"version\": \"build-10964@b29c606e2\"}, \"config_digest\": \"2874b30db323775541789cb8e3759427078d849263afe91839813790744df509\", \"hardware_key\": \"40a8c1371e06ac920aa3835029c4b3825f48de6152fdb15ad9c18f572a80b356\", \"schema_version\": 1}}, \"runtime_key\": \"qwen3.5-4b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q4km","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:49:28.828794+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"200d883c-d623-4184-b7fa-1690b549685f","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:d6ef7d6b6d33f70d69c5d1df4568941686a610cda0bde994801472a902500dd4\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 2740937888, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_identity\": {\"captured_at\": 1792140568.820449, \"fingerprint\": \"6f6d804e077b349f6a17736fedffb2759453ac4ce364dca28c7ac8b002277972\", \"identity\": {\"artifact_key\": \"6db6031159a6d3e9f72a7c5b501900908f1c70737d61a02c5c6c7eb65da853f8\", \"backend\": {\"implementation\": \"StreamContractEngine\", \"name\": \"llama_server\", \"version\": \"build-10964@b29c606e2\"}, \"config_digest\": \"2874b30db323775541789cb8e3759427078d849263afe91839813790744df509\", \"hardware_key\": \"40a8c1371e06ac920aa3835029c4b3825f48de6152fdb15ad9c18f572a80b356\", \"schema_version\": 1}}, \"runtime_key\": \"qwen3.5-4b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q4km","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:49:28.829500+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a01f79da-85ed-445f-b257-2c40d601d8fb","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:d6ef7d6b6d33f70d69c5d1df4568941686a610cda0bde994801472a902500dd4\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 2740937888, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_identity\": {\"captured_at\": 1792140568.820449, \"fingerprint\": \"6f6d804e077b349f6a17736fedffb2759453ac4ce364dca28c7ac8b002277972\", \"identity\": {\"artifact_key\": \"6db6031159a6d3e9f72a7c5b501900908f1c70737d61a02c5c6c7eb65da853f8\", \"backend\": {\"implementation\": \"StreamContractEngine\", \"name\": \"llama_server\", \"version\": \"build-10964@b29c606e2\"}, \"config_digest\": \"2874b30db323775541789cb8e3759427078d849263afe91839813790744df509\", \"hardware_key\": \"40a8c1371e06ac920aa3835029c4b3825f48de6152fdb15ad9c18f572a80b356\", \"schema_version\": 1}}, \"runtime_key\": \"qwen3.5-4b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q4km","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:49:58.970425+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"103aa706-c42d-4be1-89b0-2f542cdbfd9f","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:d6ef7d6b6d33f70d69c5d1df4568941686a610cda0bde994801472a902500dd4\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 2740937888, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_identity\": {\"captured_at\": 1792140568.820449, \"fingerprint\": \"6f6d804e077b349f6a17736fedffb2759453ac4ce364dca28c7ac8b002277972\", \"identity\": {\"artifact_key\": \"6db6031159a6d3e9f72a7c5b501900908f1c70737d61a02c5c6c7eb65da853f8\", \"backend\": {\"implementation\": \"StreamContractEngine\", \"name\": \"llama_server\", \"version\": \"build-10964@b29c606e2\"}, \"config_digest\": \"2874b30db323775541789cb8e3759427078d849263afe91839813790744df509\", \"hardware_key\": \"40a8c1371e06ac920aa3835029c4b3825f48de6152fdb15ad9c18f572a80b356\", \"schema_version\": 1}}, \"runtime_key\": \"qwen3.5-4b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q4km","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:49:58.970719+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":399.8316249344498,"error_message":null,"error_type":null,"event_id":"441b9acc-77f5-4040-b8c7-d78ea11ecfd6","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:d6ef7d6b6d33f70d69c5d1df4568941686a610cda0bde994801472a902500dd4\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 2740937888, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_identity\": {\"captured_at\": 1792140568.820449, \"fingerprint\": \"6f6d804e077b349f6a17736fedffb2759453ac4ce364dca28c7ac8b002277972\", \"identity\": {\"artifact_key\": \"6db6031159a6d3e9f72a7c5b501900908f1c70737d61a02c5c6c7eb65da853f8\", \"backend\": {\"implementation\": \"StreamContractEngine\", \"name\": \"llama_server\", \"version\": \"build-10964@b29c606e2\"}, \"config_digest\": \"2874b30db323775541789cb8e3759427078d849263afe91839813790744df509\", \"hardware_key\": \"40a8c1371e06ac920aa3835029c4b3825f48de6152fdb15ad9c18f572a80b356\", \"schema_version\": 1}}, \"runtime_key\": \"qwen3.5-4b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q4km","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:49:59.370813+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"80dd7e1b-4510-4054-a1f5-ca1dfac987bc","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:49:59.371628+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":4062.1333749732003,"error_message":null,"error_type":null,"event_id":"60ba707b-6a3d-412d-85ce-c22c856c4852","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:50:03.434153+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f144f3fd-3893-4ca1-847b-856e02f5a3a9","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:50:03.434961+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3b52896e-2330-40cd-b963-f195931940f9","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:50:03.435524+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d078071d-fcf5-4390-8a55-65ab64df2f70","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:50:38.018712+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"21af91d8-fd4c-420b-86a4-ac41c25939c5","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:50:38.018965+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":404.1801659623161,"error_message":null,"error_type":null,"event_id":"a93cf708-9754-445d-896f-06742a9be0fc","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:50:38.423331+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1fa3c915-d6b8-4fc1-b8fb-bb6108e6cff8","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:50:38.423892+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":7.4374169344082475,"error_message":null,"error_type":null,"event_id":"627039f4-4583-44c5-8f78-6a5273edd289","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:89a74e35df26090fe2c6adb16b01e32373fbcdad6966a1b23cc9b88ba35a5170\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4233677952, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 42, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"nemotron-nano-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:50:38.431614+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"64ad994b-d752-4886-8620-7e77bc715354","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:89a74e35df26090fe2c6adb16b01e32373fbcdad6966a1b23cc9b88ba35a5170\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4233677952, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 42, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"nemotron-nano-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:50:38.432076+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"bd051dad-f392-4371-b9bd-eab1ef3647d6","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:89a74e35df26090fe2c6adb16b01e32373fbcdad6966a1b23cc9b88ba35a5170\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4233677952, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 42, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"nemotron-nano-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:50:38.432696+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b20b6485-860d-40b6-881e-6f7edc5f1939","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:89a74e35df26090fe2c6adb16b01e32373fbcdad6966a1b23cc9b88ba35a5170\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4233677952, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 42, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"nemotron-nano-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:51:09.939143+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ac24cd75-a1fb-4124-8cab-656d5e3ba9d6","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:89a74e35df26090fe2c6adb16b01e32373fbcdad6966a1b23cc9b88ba35a5170\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4233677952, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 42, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"nemotron-nano-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:51:09.939392+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":402.0543750375509,"error_message":null,"error_type":null,"event_id":"964f6b75-7abc-4e5b-b7fd-cb51bf81e307","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:89a74e35df26090fe2c6adb16b01e32373fbcdad6966a1b23cc9b88ba35a5170\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4233677952, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 42, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"nemotron-nano-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"nemotron-nano-4b-q8","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:51:10.341650+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1f626028-66a1-453a-849c-c545b14a06ce","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:51:10.342185+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":5054.9803749891,"error_message":null,"error_type":null,"event_id":"2b46d6ae-0174-4091-bf28-2586ee6df00b","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:9d728f687c9bb0f47a2c5a8c0e61b6da5a7e2833ad1afcffb18c7b4eb2767886\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 5627044256, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_identity\": {\"captured_at\": 1792140675.392788, \"fingerprint\": \"774caa67f7936ab58b8ef38fcaaa85b171c37457ba52e3e3b1de0a5e845042ea\", \"identity\": {\"artifact_key\": \"cfa2c227191904b3a16a2f27994c105c51228c913536650b7e6573e80b1621ac\", \"backend\": {\"implementation\": \"StreamContractEngine\", \"name\": \"llama_server\", \"version\": \"build-10964@b29c606e2\"}, \"config_digest\": \"2874b30db323775541789cb8e3759427078d849263afe91839813790744df509\", \"hardware_key\": \"40a8c1371e06ac920aa3835029c4b3825f48de6152fdb15ad9c18f572a80b356\", \"schema_version\": 1}}, \"runtime_key\": \"qwen3.5-9b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:51:15.397517+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6d6602ab-26f6-40bf-89d0-7f546ec0abb0","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:9d728f687c9bb0f47a2c5a8c0e61b6da5a7e2833ad1afcffb18c7b4eb2767886\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 5627044256, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_identity\": {\"captured_at\": 1792140675.392788, \"fingerprint\": \"774caa67f7936ab58b8ef38fcaaa85b171c37457ba52e3e3b1de0a5e845042ea\", \"identity\": {\"artifact_key\": \"cfa2c227191904b3a16a2f27994c105c51228c913536650b7e6573e80b1621ac\", \"backend\": {\"implementation\": \"StreamContractEngine\", \"name\": \"llama_server\", \"version\": \"build-10964@b29c606e2\"}, \"config_digest\": \"2874b30db323775541789cb8e3759427078d849263afe91839813790744df509\", \"hardware_key\": \"40a8c1371e06ac920aa3835029c4b3825f48de6152fdb15ad9c18f572a80b356\", \"schema_version\": 1}}, \"runtime_key\": \"qwen3.5-9b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:51:15.398110+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6ff4acab-3d14-4d62-a599-a78d7392e935","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:9d728f687c9bb0f47a2c5a8c0e61b6da5a7e2833ad1afcffb18c7b4eb2767886\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 5627044256, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_identity\": {\"captured_at\": 1792140675.392788, \"fingerprint\": \"774caa67f7936ab58b8ef38fcaaa85b171c37457ba52e3e3b1de0a5e845042ea\", \"identity\": {\"artifact_key\": \"cfa2c227191904b3a16a2f27994c105c51228c913536650b7e6573e80b1621ac\", \"backend\": {\"implementation\": \"StreamContractEngine\", \"name\": \"llama_server\", \"version\": \"build-10964@b29c606e2\"}, \"config_digest\": \"2874b30db323775541789cb8e3759427078d849263afe91839813790744df509\", \"hardware_key\": \"40a8c1371e06ac920aa3835029c4b3825f48de6152fdb15ad9c18f572a80b356\", \"schema_version\": 1}}, \"runtime_key\": \"qwen3.5-9b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:51:15.398799+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"202b970c-0d28-4d82-8b07-5b1188b0ccd6","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:9d728f687c9bb0f47a2c5a8c0e61b6da5a7e2833ad1afcffb18c7b4eb2767886\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 5627044256, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_identity\": {\"captured_at\": 1792140675.392788, \"fingerprint\": \"774caa67f7936ab58b8ef38fcaaa85b171c37457ba52e3e3b1de0a5e845042ea\", \"identity\": {\"artifact_key\": \"cfa2c227191904b3a16a2f27994c105c51228c913536650b7e6573e80b1621ac\", \"backend\": {\"implementation\": \"StreamContractEngine\", \"name\": \"llama_server\", \"version\": \"build-10964@b29c606e2\"}, \"config_digest\": \"2874b30db323775541789cb8e3759427078d849263afe91839813790744df509\", \"hardware_key\": \"40a8c1371e06ac920aa3835029c4b3825f48de6152fdb15ad9c18f572a80b356\", \"schema_version\": 1}}, \"runtime_key\": \"qwen3.5-9b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:51:57.688854+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0c7b9dd5-cc54-4c2d-a500-d8e61f4a98d1","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:9d728f687c9bb0f47a2c5a8c0e61b6da5a7e2833ad1afcffb18c7b4eb2767886\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 5627044256, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_identity\": {\"captured_at\": 1792140675.392788, \"fingerprint\": \"774caa67f7936ab58b8ef38fcaaa85b171c37457ba52e3e3b1de0a5e845042ea\", \"identity\": {\"artifact_key\": \"cfa2c227191904b3a16a2f27994c105c51228c913536650b7e6573e80b1621ac\", \"backend\": {\"implementation\": \"StreamContractEngine\", \"name\": \"llama_server\", \"version\": \"build-10964@b29c606e2\"}, \"config_digest\": \"2874b30db323775541789cb8e3759427078d849263afe91839813790744df509\", \"hardware_key\": \"40a8c1371e06ac920aa3835029c4b3825f48de6152fdb15ad9c18f572a80b356\", \"schema_version\": 1}}, \"runtime_key\": \"qwen3.5-9b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:51:57.689197+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":402.1327079972252,"error_message":null,"error_type":null,"event_id":"088ff2e3-83c1-40b4-8570-a36b5212876e","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:9d728f687c9bb0f47a2c5a8c0e61b6da5a7e2833ad1afcffb18c7b4eb2767886\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 5627044256, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_identity\": {\"captured_at\": 1792140675.392788, \"fingerprint\": \"774caa67f7936ab58b8ef38fcaaa85b171c37457ba52e3e3b1de0a5e845042ea\", \"identity\": {\"artifact_key\": \"cfa2c227191904b3a16a2f27994c105c51228c913536650b7e6573e80b1621ac\", \"backend\": {\"implementation\": \"StreamContractEngine\", \"name\": \"llama_server\", \"version\": \"build-10964@b29c606e2\"}, \"config_digest\": \"2874b30db323775541789cb8e3759427078d849263afe91839813790744df509\", \"hardware_key\": \"40a8c1371e06ac920aa3835029c4b3825f48de6152fdb15ad9c18f572a80b356\", \"schema_version\": 1}}, \"runtime_key\": \"qwen3.5-9b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-q4km","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:51:58.091604+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"10137478-c221-4f0f-830c-c83ec32e73b6","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:51:58.092368+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":"[redacted local or credential-bearing value]","error_type":"TransportError","event_id":"1c1f0fcb-ae64-467c-a727-cb8cd8d02c36","event_type":"model.prepare.failed","metadata_json":"{\"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"failed","timestamp_utc":"2026-10-16T08:51:58.119416+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ac5ce4d1-0564-48fb-94c9-61eee80d3c63","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:51:58.119980+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":5059.716208023019,"error_message":null,"error_type":null,"event_id":"524d2f5c-8c9f-4ee7-b312-dba6b580a904","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:80688f79e2c27746cabb9c8e268a98d63ee3d732cb5c73d70fc581a1958017fc\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3803452480, \"backend\": \"llama_server\", \"quantization\": \"Q1_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"bonsai-27b-q10\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:52:03.180001+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"16fdb9f2-9ae0-49f3-8a7a-cf9adfe63650","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:80688f79e2c27746cabb9c8e268a98d63ee3d732cb5c73d70fc581a1958017fc\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3803452480, \"backend\": \"llama_server\", \"quantization\": \"Q1_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"bonsai-27b-q10\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:52:03.180569+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"27dd673d-b367-4476-bf3e-f1094dfca69c","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:80688f79e2c27746cabb9c8e268a98d63ee3d732cb5c73d70fc581a1958017fc\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3803452480, \"backend\": \"llama_server\", \"quantization\": \"Q1_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"bonsai-27b-q10\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:52:03.180942+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6e3a0897-f9e0-4ddd-ac5c-ea35eda7e0da","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:80688f79e2c27746cabb9c8e268a98d63ee3d732cb5c73d70fc581a1958017fc\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3803452480, \"backend\": \"llama_server\", \"quantization\": \"Q1_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"bonsai-27b-q10\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:53:36.412645+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d3419d9d-2343-455f-9d70-e6619abd5d2c","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:80688f79e2c27746cabb9c8e268a98d63ee3d732cb5c73d70fc581a1958017fc\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3803452480, \"backend\": \"llama_server\", \"quantization\": \"Q1_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"bonsai-27b-q10\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:53:36.412922+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":671.1296250578016,"error_message":null,"error_type":null,"event_id":"11f6ba06-9db1-4135-a1cb-efb73bd9d280","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:80688f79e2c27746cabb9c8e268a98d63ee3d732cb5c73d70fc581a1958017fc\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3803452480, \"backend\": \"llama_server\", \"quantization\": \"Q1_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"bonsai-27b-q10\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"bonsai-27b-q10","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:53:37.084318+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"bea95641-8691-4bf9-a1ec-6ab08bb24bc8","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:53:37.085097+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":"[redacted local or credential-bearing value]","error_type":"TransportError","event_id":"f0699048-f9ec-42b9-a38a-cf70ee2ae232","event_type":"model.prepare.failed","metadata_json":"{\"model_key\": \"qwen3.8-27b-iq1s\", \"model_signature\": \"sha256:model:6ec75a4d6a322071515ee47840445b23197e119cad460874f0b3212198546043\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.8-27b-iq1s","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"failed","timestamp_utc":"2026-10-16T08:53:37.117384+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0f7a077e-0a2a-480b-919f-0cb123a2f167","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\"}","model_key":"gemini-3.5-flash","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:53:37.143900+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"195d6b48-1ae9-4b77-a90f-e731da0c09c7","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:c58f103ecf1090aab4510c6708cd9446fefc4137c61ae9ca33da8cc1c5d3faf8\", \"model_key\": \"qwen3.5-0.8b-q4km\", \"model_signature\": \"sha256:model:a09b069a034e447e106034558f0b96cdcf2921e0f7acf930856f360caeb1c866\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\"}","model_key":"qwen3.5-0.8b-q4km","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:53:37.146475+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8632d9a0-0b9c-424a-9541-b6c3a18adf61","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\"}","model_key":"qwen3.5-2b-bf16","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:53:37.148084+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"166bb736-5ba1-491b-85b3-dd668446e636","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:d6ef7d6b6d33f70d69c5d1df4568941686a610cda0bde994801472a902500dd4\", \"model_key\": \"qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:4f27a24fce5077cc444082c35fa5e16579af56d41d243631a41bd0202f40b481\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\"}","model_key":"qwen3.5-4b-q4km","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:53:37.149386+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"4ebef850-71f5-49de-ba96-22e862a43296","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\"}","model_key":"qwen3.5-4b-q8","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:53:37.151128+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"61be7415-66d2-4978-b128-2cd0a2ed94d5","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:89a74e35df26090fe2c6adb16b01e32373fbcdad6966a1b23cc9b88ba35a5170\", \"model_key\": \"nemotron-nano-4b-q8\", \"model_signature\": \"sha256:model:87bd1e4e8a75eef16de10a74129a19e6752eb50b446b7650cec84e8dee69be3e\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\"}","model_key":"nemotron-nano-4b-q8","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:53:37.152678+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"11f227e7-b76e-47b8-8921-a72f8850896f","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:9d728f687c9bb0f47a2c5a8c0e61b6da5a7e2833ad1afcffb18c7b4eb2767886\", \"model_key\": \"qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:dc87c938ba5bb7540a4120ff4af7335480284b476b47415f31c793a74967ea42\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\"}","model_key":"qwen3.5-9b-q4km","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:53:37.153957+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b2ce1feb-4484-4ba8-a622-254a6d1c25b0","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:80688f79e2c27746cabb9c8e268a98d63ee3d732cb5c73d70fc581a1958017fc\", \"model_key\": \"bonsai-27b-q10\", \"model_signature\": \"sha256:model:fa0bee720418b717e2a5ae34c60712580c9e9d17c6ce62d3a2cbb639ae08d15d\", \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\"}","model_key":"bonsai-27b-q10","run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:53:37.155301+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f74c74c2-29f9-4b1c-9cf9-2c276271c42d","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 48, \"completed_cases\": 320, \"failed_cases\": 0, \"model_failures\": 2, \"planned_cases\": 320, \"run_group\": \"math-v3-complete\", \"run_id\": \"b10ad834-2bdd-409b-9073-49f0233792ea\", \"skipped_cases\": 0}","model_key":null,"run_id":"b10ad834-2bdd-409b-9073-49f0233792ea","sample_id":null,"status":null,"timestamp_utc":"2026-10-16T08:53:37.155498+00:00"}]},"b6b4123a-1d20-4c39-8644-ea00c1e5893b":{"cells":[{"benchmark_signature":"sha256:benchmark:1fa41ef05579230896c951f4e01da57ccf719647ad6f8235008243556409a7cb","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-20T10:03:44.604674+00:00","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","failure_count":0,"model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.0,"profile":"smoke","run_id":"b6b4123a-1d20-4c39-8644-ea00c1e5893b","sample_count":12,"status":"COMPLETED","task_id":"qa-abstention"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"b6b4123a-1d20-4c39-8644-ea00c1e5893b","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","family":"limite","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","parameters_b":1.0,"provider_key":"transformers","quantization":null,"run_id":"b6b4123a-1d20-4c39-8644-ea00c1e5893b","runtime_key":"transformers-local"}],"resources":[],"run":{"completed_at_utc":"2026-10-20T10:03:44.604674+00:00","created_at_utc":"2026-10-20T09:54:50.015691+00:00","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","integrity_state":"VALID","manifest_checksum":"ebc4ad6bec1c4798708fa4ce0f8383110ea6bbd3f84544c11dd2402600a4b295","profile":"smoke","run_group":"limite-violetto-qa-smoke","run_id":"b6b4123a-1d20-4c39-8644-ea00c1e5893b","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"3"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a7a3fed3-406c-4cb0-bae4-d641de44b9b6","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"qa-abstention\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"limite-1b-violetto\"], \"profile\": \"smoke\", \"run_group\": \"limite-violetto-qa-smoke\", \"run_id\": \"b6b4123a-1d20-4c39-8644-ea00c1e5893b\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"3\"}","model_key":null,"run_id":"b6b4123a-1d20-4c39-8644-ea00c1e5893b","sample_id":null,"status":null,"timestamp_utc":"2026-10-20T09:54:50.015691+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"363871c1-b1b1-4bfe-9177-29420831102f","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"provider_key\": \"transformers\", \"run_group\": \"limite-violetto-qa-smoke\", \"run_id\": \"b6b4123a-1d20-4c39-8644-ea00c1e5893b\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"b6b4123a-1d20-4c39-8644-ea00c1e5893b","sample_id":null,"status":"started","timestamp_utc":"2026-10-20T09:54:50.052159+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.4927079426124692,"error_message":null,"error_type":null,"event_id":"22358aa3-ba9e-4f7e-bc33-c32d1dfae25d","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"provider_key\": \"transformers\", \"run_group\": \"limite-violetto-qa-smoke\", \"run_id\": \"b6b4123a-1d20-4c39-8644-ea00c1e5893b\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"b6b4123a-1d20-4c39-8644-ea00c1e5893b","sample_id":null,"status":"completed","timestamp_utc":"2026-10-20T09:54:50.053667+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"9b2157cb-46f7-459c-8bc6-af2c0177c6d2","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"provider_key\": \"transformers\", \"run_group\": \"limite-violetto-qa-smoke\", \"run_id\": \"b6b4123a-1d20-4c39-8644-ea00c1e5893b\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"b6b4123a-1d20-4c39-8644-ea00c1e5893b","sample_id":null,"status":"started","timestamp_utc":"2026-10-20T09:54:50.054006+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6ec09a36-3884-4884-b2e1-c5039b875537","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:1fa41ef05579230896c951f4e01da57ccf719647ad6f8235008243556409a7cb\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"planned_cases\": 12, \"profile\": \"smoke\", \"provider_key\": \"transformers\", \"run_group\": \"limite-violetto-qa-smoke\", \"run_id\": \"b6b4123a-1d20-4c39-8644-ea00c1e5893b\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"b6b4123a-1d20-4c39-8644-ea00c1e5893b","sample_id":null,"status":"started","timestamp_utc":"2026-10-20T09:54:50.087046+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"95f231c2-d90d-4efe-bfc6-b948ed413bf3","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:1fa41ef05579230896c951f4e01da57ccf719647ad6f8235008243556409a7cb\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"profile\": \"smoke\", \"provider_key\": \"transformers\", \"run_group\": \"limite-violetto-qa-smoke\", \"run_id\": \"b6b4123a-1d20-4c39-8644-ea00c1e5893b\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"b6b4123a-1d20-4c39-8644-ea00c1e5893b","sample_id":null,"status":"completed","timestamp_utc":"2026-10-20T10:03:44.310996+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3f003e33-9239-4a7f-b711-8dc28498c950","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"provider_key\": \"transformers\", \"run_group\": \"limite-violetto-qa-smoke\", \"run_id\": \"b6b4123a-1d20-4c39-8644-ea00c1e5893b\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"b6b4123a-1d20-4c39-8644-ea00c1e5893b","sample_id":null,"status":"started","timestamp_utc":"2026-10-20T10:03:44.311216+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":287.7940410980955,"error_message":null,"error_type":null,"event_id":"9d1eae31-6358-4282-9925-35469c06bd48","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"provider_key\": \"transformers\", \"run_group\": \"limite-violetto-qa-smoke\", \"run_id\": \"b6b4123a-1d20-4c39-8644-ea00c1e5893b\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"b6b4123a-1d20-4c39-8644-ea00c1e5893b","sample_id":null,"status":"completed","timestamp_utc":"2026-10-20T10:03:44.599224+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"dc41568a-b7ba-4827-b386-ad041848b732","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:1fa41ef05579230896c951f4e01da57ccf719647ad6f8235008243556409a7cb\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"run_group\": \"limite-violetto-qa-smoke\", \"run_id\": \"b6b4123a-1d20-4c39-8644-ea00c1e5893b\"}","model_key":"limite-1b-violetto","run_id":"b6b4123a-1d20-4c39-8644-ea00c1e5893b","sample_id":null,"status":"completed","timestamp_utc":"2026-10-20T10:03:44.604402+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"4e2600ae-a697-4071-b6d2-5d8d79f6a006","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 12, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 12, \"run_group\": \"limite-violetto-qa-smoke\", \"run_id\": \"b6b4123a-1d20-4c39-8644-ea00c1e5893b\", \"skipped_cases\": 0}","model_key":null,"run_id":"b6b4123a-1d20-4c39-8644-ea00c1e5893b","sample_id":null,"status":null,"timestamp_utc":"2026-10-20T10:03:44.604674+00:00"}]},"b906e312-aa49-4397-b60e-a1c3ad6a8f87":{"cells":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T15:33:33.988827+00:00","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","failure_count":0,"model_key":"decisio-qwen3.5-4b-q4km","model_signature":"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.7012987012987013,"profile":"core","run_id":"b906e312-aa49-4397-b60e-a1c3ad6a8f87","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"b906e312-aa49-4397-b60e-a1c3ad6a8f87","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-4b-q4km","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","family":null,"model_id":"unsloth/qwen3.5-4b","model_key":"decisio-qwen3.5-4b-q4km","model_signature":"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a","parameters_b":null,"provider_key":"decisio","quantization":"Q4_K_M","run_id":"b906e312-aa49-4397-b60e-a1c3ad6a8f87","runtime_key":"decisio-local"}],"resources":[],"run":{"completed_at_utc":"2026-10-05T15:33:33.988827+00:00","created_at_utc":"2026-10-05T15:26:13.138436+00:00","git_commit":"2e7b64ca38a2db66a9d8b36910d2b2ac5da0c7ce","integrity_state":"VALID","manifest_checksum":"4e4192edeaad6b2362b7e9ecc8a102f07ceb9485bcf60f5a6d9f0512d72d1c72","profile":"core","run_group":"decisio-classification-20261005_172309","run_id":"b906e312-aa49-4397-b60e-a1c3ad6a8f87","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1720b37b-61d3-4551-a33f-44235af26697","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"intent-classification\"], \"model_keys\": [\"decisio-qwen3.5-4b-q4km\"], \"profile\": \"core\", \"run_group\": \"decisio-classification-20261005_172309\", \"run_id\": \"b906e312-aa49-4397-b60e-a1c3ad6a8f87\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"b906e312-aa49-4397-b60e-a1c3ad6a8f87","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T15:26:13.138436+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b6bb67ba-6f43-4d41-8f7f-19ced90c028a","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_172309\", \"run_id\": \"b906e312-aa49-4397-b60e-a1c3ad6a8f87\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-4b-q4km","run_id":"b906e312-aa49-4397-b60e-a1c3ad6a8f87","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T15:26:13.160744+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":3873.9600830012932,"error_message":null,"error_type":null,"event_id":"3595b9d1-676c-435a-b7b1-d662c968fc84","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_172309\", \"run_id\": \"b906e312-aa49-4397-b60e-a1c3ad6a8f87\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-4b-q4km","run_id":"b906e312-aa49-4397-b60e-a1c3ad6a8f87","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T15:26:17.035233+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d7fb27d7-2456-4ff3-ad22-ef954f14bc8b","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_172309\", \"run_id\": \"b906e312-aa49-4397-b60e-a1c3ad6a8f87\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-4b-q4km","run_id":"b906e312-aa49-4397-b60e-a1c3ad6a8f87","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T15:26:17.036128+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"fd7cd11a-a7f8-4a86-b367-4bdacf0b505f","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_172309\", \"run_id\": \"b906e312-aa49-4397-b60e-a1c3ad6a8f87\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-4b-q4km","run_id":"b906e312-aa49-4397-b60e-a1c3ad6a8f87","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T15:26:17.041875+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"407f37cb-c1bc-4786-af04-5ed08600c305","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a\", \"profile\": \"core\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_172309\", \"run_id\": \"b906e312-aa49-4397-b60e-a1c3ad6a8f87\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-4b-q4km","run_id":"b906e312-aa49-4397-b60e-a1c3ad6a8f87","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T15:33:33.980161+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d36051d7-464f-4788-8124-531c968a94cc","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_172309\", \"run_id\": \"b906e312-aa49-4397-b60e-a1c3ad6a8f87\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-4b-q4km","run_id":"b906e312-aa49-4397-b60e-a1c3ad6a8f87","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T15:33:33.980400+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.019750004867091775,"error_message":null,"error_type":null,"event_id":"d2d58170-5949-43dd-97ce-c3abbb991f03","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_172309\", \"run_id\": \"b906e312-aa49-4397-b60e-a1c3ad6a8f87\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-4b-q4km","run_id":"b906e312-aa49-4397-b60e-a1c3ad6a8f87","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T15:33:33.980573+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"181f8656-395b-4ae3-bb75-482726668819","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-4b-q4km\", \"model_signature\": \"sha256:model:8971491a3c813817b6db961a2b4f76ef242fc21d349f0b8c55e2e8bbee242f5a\", \"run_group\": \"decisio-classification-20261005_172309\", \"run_id\": \"b906e312-aa49-4397-b60e-a1c3ad6a8f87\"}","model_key":"decisio-qwen3.5-4b-q4km","run_id":"b906e312-aa49-4397-b60e-a1c3ad6a8f87","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T15:33:33.988659+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c553ca1f-9bd9-4f7f-8da2-45c987fb9480","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"decisio-classification-20261005_172309\", \"run_id\": \"b906e312-aa49-4397-b60e-a1c3ad6a8f87\", \"skipped_cases\": 0}","model_key":null,"run_id":"b906e312-aa49-4397-b60e-a1c3ad6a8f87","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T15:33:33.988827+00:00"}]},"cb837fcf-570f-4a89-bbbd-cded7a3c335b":{"cells":[{"benchmark_signature":"sha256:benchmark:bff2038c9a70a239e320793c0c579d02b35645dd0f722e6fd8e0c6a9b30b4b07","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T13:35:21.786434+00:00","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","failure_count":0,"model_key":"limite-1b-violetto","model_signature":"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.8333333333333334,"profile":"smoke","run_id":"cb837fcf-570f-4a89-bbbd-cded7a3c335b","sample_count":12,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"cb837fcf-570f-4a89-bbbd-cded7a3c335b","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","family":"limite","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65","parameters_b":1.0,"provider_key":"transformers","quantization":null,"run_id":"cb837fcf-570f-4a89-bbbd-cded7a3c335b","runtime_key":"transformers-local"}],"resources":[],"run":{"completed_at_utc":"2026-10-15T13:35:21.786434+00:00","created_at_utc":"2026-10-15T13:33:14.211567+00:00","git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","integrity_state":"VALID","manifest_checksum":"d89149e628a9f6df502a67a21a4214a93d29d3c3a812d077c3b485ef2a4a552b","profile":"smoke","run_group":"test-cap","run_id":"cb837fcf-570f-4a89-bbbd-cded7a3c335b","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"090b2123-8b62-4a3a-8950-0df50c469727","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"limite-1b-violetto\"], \"profile\": \"smoke\", \"run_group\": \"test-cap\", \"run_id\": \"cb837fcf-570f-4a89-bbbd-cded7a3c335b\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"cb837fcf-570f-4a89-bbbd-cded7a3c335b","sample_id":null,"status":null,"timestamp_utc":"2026-10-15T13:33:14.211567+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a61c30f5-4c9d-467c-af12-8b91ae72c3ca","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65\", \"provider_key\": \"transformers\", \"run_group\": \"test-cap\", \"run_id\": \"cb837fcf-570f-4a89-bbbd-cded7a3c335b\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"cb837fcf-570f-4a89-bbbd-cded7a3c335b","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T13:33:14.234795+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.3874169196933508,"error_message":null,"error_type":null,"event_id":"f03e717e-d559-4e04-8419-f9b49e03be03","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65\", \"provider_key\": \"transformers\", \"run_group\": \"test-cap\", \"run_id\": \"cb837fcf-570f-4a89-bbbd-cded7a3c335b\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"cb837fcf-570f-4a89-bbbd-cded7a3c335b","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T13:33:14.235680+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e9b49414-b541-4abc-af87-f08c51e15879","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65\", \"provider_key\": \"transformers\", \"run_group\": \"test-cap\", \"run_id\": \"cb837fcf-570f-4a89-bbbd-cded7a3c335b\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"cb837fcf-570f-4a89-bbbd-cded7a3c335b","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T13:33:14.236002+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"648c9dab-4ad2-46c6-b135-c2ec88245e76","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:bff2038c9a70a239e320793c0c579d02b35645dd0f722e6fd8e0c6a9b30b4b07\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65\", \"planned_cases\": 12, \"profile\": \"smoke\", \"provider_key\": \"transformers\", \"run_group\": \"test-cap\", \"run_id\": \"cb837fcf-570f-4a89-bbbd-cded7a3c335b\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"cb837fcf-570f-4a89-bbbd-cded7a3c335b","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T13:33:14.250147+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"062e7aac-592d-4ba0-ad8a-c21bed024f2e","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:bff2038c9a70a239e320793c0c579d02b35645dd0f722e6fd8e0c6a9b30b4b07\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65\", \"profile\": \"smoke\", \"provider_key\": \"transformers\", \"run_group\": \"test-cap\", \"run_id\": \"cb837fcf-570f-4a89-bbbd-cded7a3c335b\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"cb837fcf-570f-4a89-bbbd-cded7a3c335b","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T13:35:21.543695+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"eaf7adb0-648a-4d40-82e3-ffaed9746a3a","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65\", \"provider_key\": \"transformers\", \"run_group\": \"test-cap\", \"run_id\": \"cb837fcf-570f-4a89-bbbd-cded7a3c335b\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"cb837fcf-570f-4a89-bbbd-cded7a3c335b","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T13:35:21.543908+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":237.6590829808265,"error_message":null,"error_type":null,"event_id":"59986041-1bce-46f8-921b-89ad71c68b02","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65\", \"provider_key\": \"transformers\", \"run_group\": \"test-cap\", \"run_id\": \"cb837fcf-570f-4a89-bbbd-cded7a3c335b\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"cb837fcf-570f-4a89-bbbd-cded7a3c335b","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T13:35:21.781790+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"86c586bc-9bb0-4dce-b244-fdb43629051b","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:bff2038c9a70a239e320793c0c579d02b35645dd0f722e6fd8e0c6a9b30b4b07\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:29c7a23df4f7cd9b3388f624672a81509e07ed668585d95f62b638b3d9c6cb65\", \"run_group\": \"test-cap\", \"run_id\": \"cb837fcf-570f-4a89-bbbd-cded7a3c335b\"}","model_key":"limite-1b-violetto","run_id":"cb837fcf-570f-4a89-bbbd-cded7a3c335b","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T13:35:21.786240+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"841994f8-9fae-4939-9885-7fe8e1711505","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 12, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 12, \"run_group\": \"test-cap\", \"run_id\": \"cb837fcf-570f-4a89-bbbd-cded7a3c335b\", \"skipped_cases\": 0}","model_key":null,"run_id":"cb837fcf-570f-4a89-bbbd-cded7a3c335b","sample_id":null,"status":null,"timestamp_utc":"2026-10-15T13:35:21.786434+00:00"}]},"d445eb9f-1c5a-4781-83be-43b6eeee3a44":{"cells":[{"benchmark_signature":"sha256:benchmark:dca92e0e83499d6053d9585c0bdee4321b7ec9651563312492a659578feb94ee","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-20T20:00:36.317482+00:00","execution_signature":"sha256:execution:f4c681453ad106c8c878d601ee4a0ee6335674e81abaf3f7ba5297ce35841db0","failure_count":0,"model_key":"limite-1b-violetto-mlx","model_signature":"sha256:model:fddf01c0845257f40bbb635a92054df8eb22dacd450b827a951962e906d6513b","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.0,"profile":"smoke","run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","sample_count":10,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"safetensors","artifact_size_bytes":416,"deployment":"local","effective_model_id":"limite-1b-violetto-mlx","execution_signature":"sha256:execution:f4c681453ad106c8c878d601ee4a0ee6335674e81abaf3f7ba5297ce35841db0","family":"limite","model_id":"gabe-rbo/limite-1b-violetto-mlx","model_key":"limite-1b-violetto-mlx","model_signature":"sha256:model:fddf01c0845257f40bbb635a92054df8eb22dacd450b827a951962e906d6513b","parameters_b":1.0,"provider_key":"korgis","quantization":null,"run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:f4c681453ad106c8c878d601ee4a0ee6335674e81abaf3f7ba5297ce35841db0","model_key":"limite-1b-violetto-mlx","process_cpu_percent_avg":54.47089950027608,"process_rss_bytes_avg":96917494.20591466,"process_rss_bytes_peak":102842368.0,"sample_count":670,"sample_interval_ms":331.7098112721101,"sampling_error_count":0,"scope":"server_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":7568949248.0}],"run":{"completed_at_utc":"2026-10-20T20:00:36.317482+00:00","created_at_utc":"2026-10-20T19:56:56.412647+00:00","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","integrity_state":"VALID","manifest_checksum":"9b1ae4f12731d5f65acf6095bef57ac4d0bc7712599fd1e9bd99de7562b83d8d","profile":"smoke","run_group":"limite-mlx-korgis-smoke-test","run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","seed":42,"source_dir":null,"status":"PARTIAL","suite_id":"capability-core","suite_version":"3"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b54781b0-a2f7-4a5d-a3f7-b7edacf481f1","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"limite-1b-violetto-mlx\"], \"profile\": \"smoke\", \"run_group\": \"limite-mlx-korgis-smoke-test\", \"run_id\": \"d445eb9f-1c5a-4781-83be-43b6eeee3a44\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"3\"}","model_key":null,"run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","sample_id":null,"status":null,"timestamp_utc":"2026-10-20T19:56:56.412647+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ba243249-0832-4c1c-9b94-64af46df5534","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"limite-1b-violetto-mlx\", \"model_signature\": \"sha256:model:fddf01c0845257f40bbb635a92054df8eb22dacd450b827a951962e906d6513b\", \"provider_key\": \"korgis\", \"run_group\": \"limite-mlx-korgis-smoke-test\", \"run_id\": \"d445eb9f-1c5a-4781-83be-43b6eeee3a44\", \"runtime_key\": \"korgis-local\"}","model_key":"limite-1b-violetto-mlx","run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","sample_id":null,"status":"started","timestamp_utc":"2026-10-20T19:56:56.444881+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":114.2181670293212,"error_message":null,"error_type":null,"event_id":"20feb000-4b63-4c94-979c-29f52fcb5b82","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:f4c681453ad106c8c878d601ee4a0ee6335674e81abaf3f7ba5297ce35841db0\", \"model_key\": \"limite-1b-violetto-mlx\", \"model_signature\": \"sha256:model:fddf01c0845257f40bbb635a92054df8eb22dacd450b827a951962e906d6513b\", \"provider_key\": \"korgis\", \"run_group\": \"limite-mlx-korgis-smoke-test\", \"run_id\": \"d445eb9f-1c5a-4781-83be-43b6eeee3a44\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 416, \"backend\": \"mlx\", \"runtime_config\": {\"backend\": \"mlx-lm\", \"ctx_size\": 4096, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": 8192, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"limite-1b-violetto-mlx\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"limite-1b-violetto-mlx","run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","sample_id":null,"status":"completed","timestamp_utc":"2026-10-20T19:56:56.559861+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"634376d1-bd7d-45e8-98ac-be9b8276bece","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:f4c681453ad106c8c878d601ee4a0ee6335674e81abaf3f7ba5297ce35841db0\", \"model_key\": \"limite-1b-violetto-mlx\", \"model_signature\": \"sha256:model:fddf01c0845257f40bbb635a92054df8eb22dacd450b827a951962e906d6513b\", \"provider_key\": \"korgis\", \"run_group\": \"limite-mlx-korgis-smoke-test\", \"run_id\": \"d445eb9f-1c5a-4781-83be-43b6eeee3a44\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 416, \"backend\": \"mlx\", \"runtime_config\": {\"backend\": \"mlx-lm\", \"ctx_size\": 4096, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": 8192, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"limite-1b-violetto-mlx\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"limite-1b-violetto-mlx","run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","sample_id":null,"status":"started","timestamp_utc":"2026-10-20T19:56:56.560704+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"82c0b252-adbf-4f73-bf23-07f1eb70d852","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:dca92e0e83499d6053d9585c0bdee4321b7ec9651563312492a659578feb94ee\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:f4c681453ad106c8c878d601ee4a0ee6335674e81abaf3f7ba5297ce35841db0\", \"model_key\": \"limite-1b-violetto-mlx\", \"model_signature\": \"sha256:model:fddf01c0845257f40bbb635a92054df8eb22dacd450b827a951962e906d6513b\", \"planned_cases\": 12, \"profile\": \"smoke\", \"provider_key\": \"korgis\", \"run_group\": \"limite-mlx-korgis-smoke-test\", \"run_id\": \"d445eb9f-1c5a-4781-83be-43b6eeee3a44\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 416, \"backend\": \"mlx\", \"runtime_config\": {\"backend\": \"mlx-lm\", \"ctx_size\": 4096, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": 8192, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"limite-1b-violetto-mlx\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"limite-1b-violetto-mlx","run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","sample_id":null,"status":"started","timestamp_utc":"2026-10-20T19:56:56.589634+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"733a5a55-e07a-4c20-ab7e-41cd1c20237b","event_type":"capability.aborted","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:dca92e0e83499d6053d9585c0bdee4321b7ec9651563312492a659578feb94ee\", \"capability_id\": \"mathematical-reasoning\", \"consecutive_failures\": 10, \"execution_signature\": \"sha256:execution:f4c681453ad106c8c878d601ee4a0ee6335674e81abaf3f7ba5297ce35841db0\", \"model_key\": \"limite-1b-violetto-mlx\", \"model_signature\": \"sha256:model:fddf01c0845257f40bbb635a92054df8eb22dacd450b827a951962e906d6513b\", \"profile\": \"smoke\", \"provider_key\": \"korgis\", \"reason\": \"Circuit breaker tripped after 10 consecutive invalid cases\", \"run_group\": \"limite-mlx-korgis-smoke-test\", \"run_id\": \"d445eb9f-1c5a-4781-83be-43b6eeee3a44\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 416, \"backend\": \"mlx\", \"runtime_config\": {\"backend\": \"mlx-lm\", \"ctx_size\": 4096, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": 8192, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"limite-1b-violetto-mlx\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"limite-1b-violetto-mlx","run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","sample_id":null,"status":"aborted","timestamp_utc":"2026-10-20T20:00:36.310450+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b41adbc8-f749-4618-859a-95936d3c8a7c","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:f4c681453ad106c8c878d601ee4a0ee6335674e81abaf3f7ba5297ce35841db0\", \"model_key\": \"limite-1b-violetto-mlx\", \"model_signature\": \"sha256:model:fddf01c0845257f40bbb635a92054df8eb22dacd450b827a951962e906d6513b\", \"provider_key\": \"korgis\", \"run_group\": \"limite-mlx-korgis-smoke-test\", \"run_id\": \"d445eb9f-1c5a-4781-83be-43b6eeee3a44\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 416, \"backend\": \"mlx\", \"runtime_config\": {\"backend\": \"mlx-lm\", \"ctx_size\": 4096, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": 8192, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"limite-1b-violetto-mlx\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"limite-1b-violetto-mlx","run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","sample_id":null,"status":"started","timestamp_utc":"2026-10-20T20:00:36.310736+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":3.839749959297478,"error_message":null,"error_type":null,"event_id":"e4a9fd39-4dec-4a13-b385-1757b33c29b2","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:f4c681453ad106c8c878d601ee4a0ee6335674e81abaf3f7ba5297ce35841db0\", \"model_key\": \"limite-1b-violetto-mlx\", \"model_signature\": \"sha256:model:fddf01c0845257f40bbb635a92054df8eb22dacd450b827a951962e906d6513b\", \"provider_key\": \"korgis\", \"run_group\": \"limite-mlx-korgis-smoke-test\", \"run_id\": \"d445eb9f-1c5a-4781-83be-43b6eeee3a44\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 416, \"backend\": \"mlx\", \"runtime_config\": {\"backend\": \"mlx-lm\", \"ctx_size\": 4096, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": 8192, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"limite-1b-violetto-mlx\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"limite-1b-violetto-mlx","run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","sample_id":null,"status":"completed","timestamp_utc":"2026-10-20T20:00:36.314773+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"38a85485-459d-4670-a91d-ecbdeb724fed","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:dca92e0e83499d6053d9585c0bdee4321b7ec9651563312492a659578feb94ee\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:f4c681453ad106c8c878d601ee4a0ee6335674e81abaf3f7ba5297ce35841db0\", \"model_key\": \"limite-1b-violetto-mlx\", \"model_signature\": \"sha256:model:fddf01c0845257f40bbb635a92054df8eb22dacd450b827a951962e906d6513b\", \"run_group\": \"limite-mlx-korgis-smoke-test\", \"run_id\": \"d445eb9f-1c5a-4781-83be-43b6eeee3a44\"}","model_key":"limite-1b-violetto-mlx","run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","sample_id":null,"status":"completed","timestamp_utc":"2026-10-20T20:00:36.317253+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"730861e6-cb33-49c5-b374-6f97fddd8faa","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 10, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 12, \"run_group\": \"limite-mlx-korgis-smoke-test\", \"run_id\": \"d445eb9f-1c5a-4781-83be-43b6eeee3a44\", \"skipped_cases\": 0}","model_key":null,"run_id":"d445eb9f-1c5a-4781-83be-43b6eeee3a44","sample_id":null,"status":null,"timestamp_utc":"2026-10-20T20:00:36.317482+00:00"}]},"e0c2dc7c-ff6a-4895-bb32-54ec8a2448d6":{"cells":[{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-16T08:42:51.527523+00:00","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","failure_count":0,"model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.8,"profile":"core","run_id":"e0c2dc7c-ff6a-4895-bb32-54ec8a2448d6","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"e0c2dc7c-ff6a-4895-bb32-54ec8a2448d6","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"local","effective_model_id":"paradigma-inc/limite-1b-violetto","execution_signature":"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2","family":"limite","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto","model_signature":"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69","parameters_b":1.0,"provider_key":"transformers","quantization":null,"run_id":"e0c2dc7c-ff6a-4895-bb32-54ec8a2448d6","runtime_key":"transformers-local"}],"resources":[],"run":{"completed_at_utc":"2026-10-16T08:42:51.527523+00:00","created_at_utc":"2026-10-16T08:31:46.449749+00:00","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","integrity_state":"VALID","manifest_checksum":"87e8dbe589a637ed618c88bccc6decc829d3807997edc124c895b4393e4fded8","profile":"core","run_group":"limite-mathematics","run_id":"e0c2dc7c-ff6a-4895-bb32-54ec8a2448d6","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"3"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b619db25-9dde-4943-9186-654916c284b5","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning-v2\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"limite-1b-violetto\"], \"profile\": \"core\", \"run_group\": \"limite-mathematics\", \"run_id\": \"e0c2dc7c-ff6a-4895-bb32-54ec8a2448d6\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"3\"}","model_key":null,"run_id":"e0c2dc7c-ff6a-4895-bb32-54ec8a2448d6","sample_id":null,"status":null,"timestamp_utc":"2026-10-16T08:31:46.449749+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2ac14a74-703f-4826-a30f-71b1aeefd23b","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"provider_key\": \"transformers\", \"run_group\": \"limite-mathematics\", \"run_id\": \"e0c2dc7c-ff6a-4895-bb32-54ec8a2448d6\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"e0c2dc7c-ff6a-4895-bb32-54ec8a2448d6","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:31:46.473244+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.27845799922943115,"error_message":null,"error_type":null,"event_id":"fb9c4087-1e2d-41b5-b403-e80d174ca843","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"provider_key\": \"transformers\", \"run_group\": \"limite-mathematics\", \"run_id\": \"e0c2dc7c-ff6a-4895-bb32-54ec8a2448d6\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"e0c2dc7c-ff6a-4895-bb32-54ec8a2448d6","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:31:46.474062+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7f542dd2-63ac-452a-833f-bb54d615e51a","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning-v2\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"provider_key\": \"transformers\", \"run_group\": \"limite-mathematics\", \"run_id\": \"e0c2dc7c-ff6a-4895-bb32-54ec8a2448d6\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"e0c2dc7c-ff6a-4895-bb32-54ec8a2448d6","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:31:46.474275+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c475f190-f936-4580-bb14-d5186c660424","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:738d15e8d280a7feb3ee7f9fda46cfc9b6c4515af2236f69dde793dbdbe30496\", \"capability_id\": \"mathematical-reasoning-v2\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"transformers\", \"run_group\": \"limite-mathematics\", \"run_id\": \"e0c2dc7c-ff6a-4895-bb32-54ec8a2448d6\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"e0c2dc7c-ff6a-4895-bb32-54ec8a2448d6","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:31:46.489010+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"33d23f70-4f0a-4882-9d07-5647b7e133e9","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:738d15e8d280a7feb3ee7f9fda46cfc9b6c4515af2236f69dde793dbdbe30496\", \"capability_id\": \"mathematical-reasoning-v2\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"profile\": \"core\", \"provider_key\": \"transformers\", \"run_group\": \"limite-mathematics\", \"run_id\": \"e0c2dc7c-ff6a-4895-bb32-54ec8a2448d6\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"e0c2dc7c-ff6a-4895-bb32-54ec8a2448d6","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:42:51.228838+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"05a8921d-6bdc-4ba3-a40a-6735b8334d6c","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"provider_key\": \"transformers\", \"run_group\": \"limite-mathematics\", \"run_id\": \"e0c2dc7c-ff6a-4895-bb32-54ec8a2448d6\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"e0c2dc7c-ff6a-4895-bb32-54ec8a2448d6","sample_id":null,"status":"started","timestamp_utc":"2026-10-16T08:42:51.229128+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":290.27658293489367,"error_message":null,"error_type":null,"event_id":"42d18b1b-9cad-45ad-8e23-34c1c07952b3","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"provider_key\": \"transformers\", \"run_group\": \"limite-mathematics\", \"run_id\": \"e0c2dc7c-ff6a-4895-bb32-54ec8a2448d6\", \"runtime_key\": \"transformers-local\"}","model_key":"limite-1b-violetto","run_id":"e0c2dc7c-ff6a-4895-bb32-54ec8a2448d6","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:42:51.519692+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"da7e3f6e-ab47-4497-8a5c-13d2a2adc049","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:738d15e8d280a7feb3ee7f9fda46cfc9b6c4515af2236f69dde793dbdbe30496\", \"capability_id\": \"mathematical-reasoning-v2\", \"execution_signature\": \"sha256:execution:b396c8355686dfca8abc1693b1f4c50dc70481ed9efa64d96823e45032f5e7c2\", \"model_key\": \"limite-1b-violetto\", \"model_signature\": \"sha256:model:35c8a917b04bf3560eae96db44a09e9b12f2edbe19cee9e9d2c0a351a42b9a69\", \"run_group\": \"limite-mathematics\", \"run_id\": \"e0c2dc7c-ff6a-4895-bb32-54ec8a2448d6\"}","model_key":"limite-1b-violetto","run_id":"e0c2dc7c-ff6a-4895-bb32-54ec8a2448d6","sample_id":null,"status":"completed","timestamp_utc":"2026-10-16T08:42:51.527278+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"9485603f-443d-461c-be77-a8e4a198cafe","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 40, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 40, \"run_group\": \"limite-mathematics\", \"run_id\": \"e0c2dc7c-ff6a-4895-bb32-54ec8a2448d6\", \"skipped_cases\": 0}","model_key":null,"run_id":"e0c2dc7c-ff6a-4895-bb32-54ec8a2448d6","sample_id":null,"status":null,"timestamp_utc":"2026-10-16T08:42:51.527523+00:00"}]},"e68754ab-25e6-4449-b06a-66dbc311d9e5":{"cells":[{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T15:45:57.052259+00:00","execution_signature":"sha256:execution:d26122294632fb1ef7e62445eece13d61030562e5a3d7a64b60606c3c12ff3be","failure_count":0,"model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:8dcf43803761fd5570f637e166b655565b6c638e51a6a96848f58f6e6699a734","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.3,"profile":"core","run_id":"e68754ab-25e6-4449-b06a-66dbc311d9e5","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"e68754ab-25e6-4449-b06a-66dbc311d9e5","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":2600224352,"deployment":"local","effective_model_id":"spark-x2.5-4b-q4km","execution_signature":"sha256:execution:d26122294632fb1ef7e62445eece13d61030562e5a3d7a64b60606c3c12ff3be","family":"spark-x2.5","model_id":"XHToken/Spark-X2.5-4B","model_key":"spark-x2.5-4b-q4km","model_signature":"sha256:model:8dcf43803761fd5570f637e166b655565b6c638e51a6a96848f58f6e6699a734","parameters_b":4.0,"provider_key":"korgis","quantization":"Q4_K_M","run_id":"e68754ab-25e6-4449-b06a-66dbc311d9e5","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning","execution_signature":"sha256:execution:d26122294632fb1ef7e62445eece13d61030562e5a3d7a64b60606c3c12ff3be","model_key":"spark-x2.5-4b-q4km","process_cpu_percent_avg":6.338877896022156,"process_rss_bytes_avg":3725684701.8666673,"process_rss_bytes_peak":4221812736.0,"sample_count":122,"sample_interval_ms":266.60380743026815,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":8769241088.0}],"run":{"completed_at_utc":"2026-10-15T15:45:57.052259+00:00","created_at_utc":"2026-10-15T15:45:30.462030+00:00","git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","integrity_state":"VALID","manifest_checksum":"8da0a45f690eeef9eaf85c2308d6ebedfb4c3456278fb7a1e0a72ce8e6696cee","profile":"core","run_group":"math-v3-core-spark-x2.5-4b-q4km","run_id":"e68754ab-25e6-4449-b06a-66dbc311d9e5","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"3"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"23c86f8a-f41f-404f-8e70-81b6f3a09bdd","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"spark-x2.5-4b-q4km\"], \"profile\": \"core\", \"run_group\": \"math-v3-core-spark-x2.5-4b-q4km\", \"run_id\": \"e68754ab-25e6-4449-b06a-66dbc311d9e5\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"3\"}","model_key":null,"run_id":"e68754ab-25e6-4449-b06a-66dbc311d9e5","sample_id":null,"status":null,"timestamp_utc":"2026-10-15T15:45:30.462030+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e0b8e1ce-f051-463c-b8e5-9ebeb4beb1ec","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:8dcf43803761fd5570f637e166b655565b6c638e51a6a96848f58f6e6699a734\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-core-spark-x2.5-4b-q4km\", \"run_id\": \"e68754ab-25e6-4449-b06a-66dbc311d9e5\", \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"e68754ab-25e6-4449-b06a-66dbc311d9e5","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T15:45:30.482675+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2095.6248329021037,"error_message":null,"error_type":null,"event_id":"a86280ec-85fe-4538-9b6a-eaa377a6deb8","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:d26122294632fb1ef7e62445eece13d61030562e5a3d7a64b60606c3c12ff3be\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:8dcf43803761fd5570f637e166b655565b6c638e51a6a96848f58f6e6699a734\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-core-spark-x2.5-4b-q4km\", \"run_id\": \"e68754ab-25e6-4449-b06a-66dbc311d9e5\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 2600224352, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_identity\": {\"captured_at\": 1792079132.537235, \"fingerprint\": \"56ae89fffcdccd2f78df03421e03a40191743c538a4cc4ba90420e3da7618ce4\", \"identity\": {\"artifact_key\": \"6c2fa9e199265e6323e4f6c498a4e82a50c601d3a25a5b68808de26ce1e7a180\", \"backend\": {\"implementation\": \"StreamContractEngine\", \"name\": \"llama_server\", \"version\": \"build-10964@b29c606e2\"}, \"config_digest\": \"f054d69ec626b099167c5283c4a81730ee8055dad313462e5b7be616941643a4\", \"hardware_key\": \"40a8c1371e06ac920aa3835029c4b3825f48de6152fdb15ad9c18f572a80b356\", \"schema_version\": 1}}, \"runtime_key\": \"spark-x2.5-4b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"e68754ab-25e6-4449-b06a-66dbc311d9e5","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T15:45:32.578709+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"9b325e8a-b2de-42b5-b6d1-650f31397c9e","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:d26122294632fb1ef7e62445eece13d61030562e5a3d7a64b60606c3c12ff3be\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:8dcf43803761fd5570f637e166b655565b6c638e51a6a96848f58f6e6699a734\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-core-spark-x2.5-4b-q4km\", \"run_id\": \"e68754ab-25e6-4449-b06a-66dbc311d9e5\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 2600224352, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_identity\": {\"captured_at\": 1792079132.537235, \"fingerprint\": \"56ae89fffcdccd2f78df03421e03a40191743c538a4cc4ba90420e3da7618ce4\", \"identity\": {\"artifact_key\": \"6c2fa9e199265e6323e4f6c498a4e82a50c601d3a25a5b68808de26ce1e7a180\", \"backend\": {\"implementation\": \"StreamContractEngine\", \"name\": \"llama_server\", \"version\": \"build-10964@b29c606e2\"}, \"config_digest\": \"f054d69ec626b099167c5283c4a81730ee8055dad313462e5b7be616941643a4\", \"hardware_key\": \"40a8c1371e06ac920aa3835029c4b3825f48de6152fdb15ad9c18f572a80b356\", \"schema_version\": 1}}, \"runtime_key\": \"spark-x2.5-4b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"e68754ab-25e6-4449-b06a-66dbc311d9e5","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T15:45:32.578964+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"eb26d804-c448-467e-aa88-2bc2395328aa","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:d26122294632fb1ef7e62445eece13d61030562e5a3d7a64b60606c3c12ff3be\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:8dcf43803761fd5570f637e166b655565b6c638e51a6a96848f58f6e6699a734\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-core-spark-x2.5-4b-q4km\", \"run_id\": \"e68754ab-25e6-4449-b06a-66dbc311d9e5\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 2600224352, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_identity\": {\"captured_at\": 1792079132.537235, \"fingerprint\": \"56ae89fffcdccd2f78df03421e03a40191743c538a4cc4ba90420e3da7618ce4\", \"identity\": {\"artifact_key\": \"6c2fa9e199265e6323e4f6c498a4e82a50c601d3a25a5b68808de26ce1e7a180\", \"backend\": {\"implementation\": \"StreamContractEngine\", \"name\": \"llama_server\", \"version\": \"build-10964@b29c606e2\"}, \"config_digest\": \"f054d69ec626b099167c5283c4a81730ee8055dad313462e5b7be616941643a4\", \"hardware_key\": \"40a8c1371e06ac920aa3835029c4b3825f48de6152fdb15ad9c18f572a80b356\", \"schema_version\": 1}}, \"runtime_key\": \"spark-x2.5-4b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"e68754ab-25e6-4449-b06a-66dbc311d9e5","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T15:45:32.596357+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"71619a1e-0470-4db7-a809-296ffe1d00e2","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:d26122294632fb1ef7e62445eece13d61030562e5a3d7a64b60606c3c12ff3be\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:8dcf43803761fd5570f637e166b655565b6c638e51a6a96848f58f6e6699a734\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-core-spark-x2.5-4b-q4km\", \"run_id\": \"e68754ab-25e6-4449-b06a-66dbc311d9e5\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 2600224352, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_identity\": {\"captured_at\": 1792079132.537235, \"fingerprint\": \"56ae89fffcdccd2f78df03421e03a40191743c538a4cc4ba90420e3da7618ce4\", \"identity\": {\"artifact_key\": \"6c2fa9e199265e6323e4f6c498a4e82a50c601d3a25a5b68808de26ce1e7a180\", \"backend\": {\"implementation\": \"StreamContractEngine\", \"name\": \"llama_server\", \"version\": \"build-10964@b29c606e2\"}, \"config_digest\": \"f054d69ec626b099167c5283c4a81730ee8055dad313462e5b7be616941643a4\", \"hardware_key\": \"40a8c1371e06ac920aa3835029c4b3825f48de6152fdb15ad9c18f572a80b356\", \"schema_version\": 1}}, \"runtime_key\": \"spark-x2.5-4b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"e68754ab-25e6-4449-b06a-66dbc311d9e5","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T15:45:56.973369+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"29d01724-3674-41bb-8589-4f4e6173d7a6","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:d26122294632fb1ef7e62445eece13d61030562e5a3d7a64b60606c3c12ff3be\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:8dcf43803761fd5570f637e166b655565b6c638e51a6a96848f58f6e6699a734\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-core-spark-x2.5-4b-q4km\", \"run_id\": \"e68754ab-25e6-4449-b06a-66dbc311d9e5\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 2600224352, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_identity\": {\"captured_at\": 1792079132.537235, \"fingerprint\": \"56ae89fffcdccd2f78df03421e03a40191743c538a4cc4ba90420e3da7618ce4\", \"identity\": {\"artifact_key\": \"6c2fa9e199265e6323e4f6c498a4e82a50c601d3a25a5b68808de26ce1e7a180\", \"backend\": {\"implementation\": \"StreamContractEngine\", \"name\": \"llama_server\", \"version\": \"build-10964@b29c606e2\"}, \"config_digest\": \"f054d69ec626b099167c5283c4a81730ee8055dad313462e5b7be616941643a4\", \"hardware_key\": \"40a8c1371e06ac920aa3835029c4b3825f48de6152fdb15ad9c18f572a80b356\", \"schema_version\": 1}}, \"runtime_key\": \"spark-x2.5-4b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"e68754ab-25e6-4449-b06a-66dbc311d9e5","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T15:45:56.973571+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":75.74937504250556,"error_message":null,"error_type":null,"event_id":"d8044d31-9e90-43ff-9224-9ab9fcb0d780","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:d26122294632fb1ef7e62445eece13d61030562e5a3d7a64b60606c3c12ff3be\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:8dcf43803761fd5570f637e166b655565b6c638e51a6a96848f58f6e6699a734\", \"provider_key\": \"korgis\", \"run_group\": \"math-v3-core-spark-x2.5-4b-q4km\", \"run_id\": \"e68754ab-25e6-4449-b06a-66dbc311d9e5\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 2600224352, \"backend\": \"llama_server\", \"quantization\": \"Q4_K_M\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_identity\": {\"captured_at\": 1792079132.537235, \"fingerprint\": \"56ae89fffcdccd2f78df03421e03a40191743c538a4cc4ba90420e3da7618ce4\", \"identity\": {\"artifact_key\": \"6c2fa9e199265e6323e4f6c498a4e82a50c601d3a25a5b68808de26ce1e7a180\", \"backend\": {\"implementation\": \"StreamContractEngine\", \"name\": \"llama_server\", \"version\": \"build-10964@b29c606e2\"}, \"config_digest\": \"f054d69ec626b099167c5283c4a81730ee8055dad313462e5b7be616941643a4\", \"hardware_key\": \"40a8c1371e06ac920aa3835029c4b3825f48de6152fdb15ad9c18f572a80b356\", \"schema_version\": 1}}, \"runtime_key\": \"spark-x2.5-4b-q4km\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"spark-x2.5-4b-q4km","run_id":"e68754ab-25e6-4449-b06a-66dbc311d9e5","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T15:45:57.049478+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"5ec6b7f4-6e53-4ea7-9106-130013e029cc","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:d26122294632fb1ef7e62445eece13d61030562e5a3d7a64b60606c3c12ff3be\", \"model_key\": \"spark-x2.5-4b-q4km\", \"model_signature\": \"sha256:model:8dcf43803761fd5570f637e166b655565b6c638e51a6a96848f58f6e6699a734\", \"run_group\": \"math-v3-core-spark-x2.5-4b-q4km\", \"run_id\": \"e68754ab-25e6-4449-b06a-66dbc311d9e5\"}","model_key":"spark-x2.5-4b-q4km","run_id":"e68754ab-25e6-4449-b06a-66dbc311d9e5","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T15:45:57.052130+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"5bd925ac-c42a-46f3-a42c-5e583ad8cc10","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 40, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 40, \"run_group\": \"math-v3-core-spark-x2.5-4b-q4km\", \"run_id\": \"e68754ab-25e6-4449-b06a-66dbc311d9e5\", \"skipped_cases\": 0}","model_key":null,"run_id":"e68754ab-25e6-4449-b06a-66dbc311d9e5","sample_id":null,"status":null,"timestamp_utc":"2026-10-15T15:45:57.052259+00:00"}]},"e9c4ec37-433f-4b06-b397-9d9b3a4f7480":{"cells":[{"benchmark_signature":"sha256:benchmark:718e761e4ba3410c289a5bba91a6474827caaa8da30e644b3df925ba4d16fbfc","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-05T16:17:35.400024+00:00","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","failure_count":0,"model_key":"decisio-qwen3.5-9b-q4km","model_signature":"sha256:model:ae0099eeb8925bc828c92072ba2f486986899854d2d3de76e5689cd383c783e3","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6623376623376623,"profile":"smoke","run_id":"e9c4ec37-433f-4b06-b397-9d9b3a4f7480","sample_count":77,"status":"COMPLETED","task_id":"intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"e9c4ec37-433f-4b06-b397-9d9b3a4f7480","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"qwen3.5-9b-q4km","execution_signature":"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4","family":null,"model_id":"qwen/qwen3.5-9b","model_key":"decisio-qwen3.5-9b-q4km","model_signature":"sha256:model:ae0099eeb8925bc828c92072ba2f486986899854d2d3de76e5689cd383c783e3","parameters_b":null,"provider_key":"decisio","quantization":"Q4_K_M","run_id":"e9c4ec37-433f-4b06-b397-9d9b3a4f7480","runtime_key":"decisio-local"}],"resources":[],"run":{"completed_at_utc":"2026-10-05T16:17:35.400024+00:00","created_at_utc":"2026-10-05T16:11:02.917234+00:00","git_commit":"0065acc5513f65c3e8bbdf6899cf55e286dfe938","integrity_state":"VALID","manifest_checksum":"4bd45b534c70c58b0de87a62ef72f68851b7a131b5d22410f04fd4532c16581d","profile":"smoke","run_group":"decisio-classification-20261005_181101","run_id":"e9c4ec37-433f-4b06-b397-9d9b3a4f7480","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b1f947c4-66a6-4576-b7ed-ec4a35e8bbf6","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"intent-classification\"], \"model_keys\": [\"decisio-qwen3.5-9b-q4km\"], \"profile\": \"smoke\", \"run_group\": \"decisio-classification-20261005_181101\", \"run_id\": \"e9c4ec37-433f-4b06-b397-9d9b3a4f7480\", \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"e9c4ec37-433f-4b06-b397-9d9b3a4f7480","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T16:11:02.917234+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c2a791dc-3039-41a9-a84f-45756998fbe5","event_type":"model.prepare.started","metadata_json":"{\"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:ae0099eeb8925bc828c92072ba2f486986899854d2d3de76e5689cd383c783e3\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_181101\", \"run_id\": \"e9c4ec37-433f-4b06-b397-9d9b3a4f7480\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-9b-q4km","run_id":"e9c4ec37-433f-4b06-b397-9d9b3a4f7480","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T16:11:02.931121+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":7180.116999996244,"error_message":null,"error_type":null,"event_id":"f5895e0f-a7d4-4b03-8f86-c313f4d7d36d","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:ae0099eeb8925bc828c92072ba2f486986899854d2d3de76e5689cd383c783e3\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_181101\", \"run_id\": \"e9c4ec37-433f-4b06-b397-9d9b3a4f7480\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-9b-q4km","run_id":"e9c4ec37-433f-4b06-b397-9d9b3a4f7480","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T16:11:10.111784+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0a50e2de-7a7b-4d0f-94a9-98c62f46ffb0","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:ae0099eeb8925bc828c92072ba2f486986899854d2d3de76e5689cd383c783e3\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_181101\", \"run_id\": \"e9c4ec37-433f-4b06-b397-9d9b3a4f7480\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-9b-q4km","run_id":"e9c4ec37-433f-4b06-b397-9d9b3a4f7480","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T16:11:10.112570+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"9bb8ae68-6824-4c5b-b878-f42699646e57","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:718e761e4ba3410c289a5bba91a6474827caaa8da30e644b3df925ba4d16fbfc\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:ae0099eeb8925bc828c92072ba2f486986899854d2d3de76e5689cd383c783e3\", \"planned_cases\": 77, \"profile\": \"smoke\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_181101\", \"run_id\": \"e9c4ec37-433f-4b06-b397-9d9b3a4f7480\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-9b-q4km","run_id":"e9c4ec37-433f-4b06-b397-9d9b3a4f7480","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T16:11:10.120070+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d867cbd6-abac-480e-804d-f0d75e9882c2","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:718e761e4ba3410c289a5bba91a6474827caaa8da30e644b3df925ba4d16fbfc\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:ae0099eeb8925bc828c92072ba2f486986899854d2d3de76e5689cd383c783e3\", \"profile\": \"smoke\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_181101\", \"run_id\": \"e9c4ec37-433f-4b06-b397-9d9b3a4f7480\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-9b-q4km","run_id":"e9c4ec37-433f-4b06-b397-9d9b3a4f7480","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T16:17:35.393314+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2a439154-a112-4906-b145-f1f1a52eeb1c","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:ae0099eeb8925bc828c92072ba2f486986899854d2d3de76e5689cd383c783e3\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_181101\", \"run_id\": \"e9c4ec37-433f-4b06-b397-9d9b3a4f7480\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-9b-q4km","run_id":"e9c4ec37-433f-4b06-b397-9d9b3a4f7480","sample_id":null,"status":"started","timestamp_utc":"2026-10-05T16:17:35.393553+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.017583006410859525,"error_message":null,"error_type":null,"event_id":"e253d95e-d357-4801-9806-86f6938a4f48","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:ae0099eeb8925bc828c92072ba2f486986899854d2d3de76e5689cd383c783e3\", \"provider_key\": \"decisio\", \"run_group\": \"decisio-classification-20261005_181101\", \"run_id\": \"e9c4ec37-433f-4b06-b397-9d9b3a4f7480\", \"runtime_key\": \"decisio-local\"}","model_key":"decisio-qwen3.5-9b-q4km","run_id":"e9c4ec37-433f-4b06-b397-9d9b3a4f7480","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T16:17:35.393793+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"71617369-b962-4ed6-ab5b-b9c2eeb4275f","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:718e761e4ba3410c289a5bba91a6474827caaa8da30e644b3df925ba4d16fbfc\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:00b5980c85691f196adbf5161683304dd8c69e0e104a56c9f609fc176f2673d4\", \"model_key\": \"decisio-qwen3.5-9b-q4km\", \"model_signature\": \"sha256:model:ae0099eeb8925bc828c92072ba2f486986899854d2d3de76e5689cd383c783e3\", \"run_group\": \"decisio-classification-20261005_181101\", \"run_id\": \"e9c4ec37-433f-4b06-b397-9d9b3a4f7480\"}","model_key":"decisio-qwen3.5-9b-q4km","run_id":"e9c4ec37-433f-4b06-b397-9d9b3a4f7480","sample_id":null,"status":"completed","timestamp_utc":"2026-10-05T16:17:35.399847+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"843e893e-8d5e-4b7e-9c57-533c4790d457","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 77, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 77, \"run_group\": \"decisio-classification-20261005_181101\", \"run_id\": \"e9c4ec37-433f-4b06-b397-9d9b3a4f7480\", \"skipped_cases\": 0}","model_key":null,"run_id":"e9c4ec37-433f-4b06-b397-9d9b3a4f7480","sample_id":null,"status":null,"timestamp_utc":"2026-10-05T16:17:35.400024+00:00"}]},"ea0fd04b-aae3-4020-ac46-99d066560ecb":{"cells":[{"benchmark_signature":"sha256:benchmark:15d8a1158c13039560913694caad4a09b3d371b1f70c0483756bbbb70603d9bf","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-20T12:33:29.825206+00:00","execution_signature":"sha256:execution:1b1fd8e53d6820b3449bce86d39dff358b1d6b5726cab7289358ad7888484aa8","failure_count":0,"model_key":"limite-1b-violetto-q80","model_signature":"sha256:model:5c5de7987a023b955d7b46893217f72c1c6b1523f722a2097fec543baf5ff73d","practical_delta":0.03,"primary_metric":"accuracy","primary_value":null,"profile":"smoke","run_id":"ea0fd04b-aae3-4020-ac46-99d066560ecb","sample_count":0,"status":"PARTIAL","task_id":"intent-classification"},{"benchmark_signature":"sha256:benchmark:5b229f97176a163f87003f87f38f0742afa7bfa9ef34614ba8bfa3dc68444556","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-20T12:33:29.825206+00:00","execution_signature":"sha256:execution:1b1fd8e53d6820b3449bce86d39dff358b1d6b5726cab7289358ad7888484aa8","failure_count":0,"model_key":"limite-1b-violetto-q80","model_signature":"sha256:model:5c5de7987a023b955d7b46893217f72c1c6b1523f722a2097fec543baf5ff73d","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":null,"profile":"smoke","run_id":"ea0fd04b-aae3-4020-ac46-99d066560ecb","sample_count":0,"status":"PARTIAL","task_id":"qa-abstention"},{"benchmark_signature":"sha256:benchmark:a1c425ecab7464c09d83d85bf8d26737f6a118eb196743746169f1c8dfb8c3d2","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-20T12:33:29.825206+00:00","execution_signature":"sha256:execution:1b1fd8e53d6820b3449bce86d39dff358b1d6b5726cab7289358ad7888484aa8","failure_count":0,"model_key":"limite-1b-violetto-q80","model_signature":"sha256:model:5c5de7987a023b955d7b46893217f72c1c6b1523f722a2097fec543baf5ff73d","practical_delta":0.05,"primary_metric":"exact_match","primary_value":null,"profile":"smoke","run_id":"ea0fd04b-aae3-4020-ac46-99d066560ecb","sample_count":0,"status":"PARTIAL","task_id":"structured-output"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"ea0fd04b-aae3-4020-ac46-99d066560ecb","runtime_json":"{}","sweep_id":null},"failure_summary":[{"count":1,"error_type":"TransportError","event_type":"model.prepare.failed","first_at":"2026-10-20T12:33:29.824779+00:00","last_at":"2026-10-20T12:33:29.824779+00:00"}],"models":[{"artifact_format":"gguf","artifact_size_bytes":null,"deployment":"local","effective_model_id":"limite-1b-violetto-q80","execution_signature":"sha256:execution:1b1fd8e53d6820b3449bce86d39dff358b1d6b5726cab7289358ad7888484aa8","family":"limite","model_id":"paradigma-inc/limite-1b-violetto","model_key":"limite-1b-violetto-q80","model_signature":"sha256:model:5c5de7987a023b955d7b46893217f72c1c6b1523f722a2097fec543baf5ff73d","parameters_b":1.0,"provider_key":"korgis","quantization":"Q8_0","run_id":"ea0fd04b-aae3-4020-ac46-99d066560ecb","runtime_key":"korgis-local"}],"resources":[],"run":{"completed_at_utc":"2026-10-20T12:33:29.825206+00:00","created_at_utc":"2026-10-20T12:33:29.776734+00:00","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","integrity_state":"VALID","manifest_checksum":"b48449dfb5de77769f39c1c03140956377f53f95d7883cdfb75dcfcf613fc6c4","profile":"smoke","run_group":"limite-violetto-q80-smoke","run_id":"ea0fd04b-aae3-4020-ac46-99d066560ecb","seed":42,"source_dir":null,"status":"PARTIAL","suite_id":"capability-core","suite_version":"3"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ec267585-95c1-4617-a395-c49a8c463798","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"intent-classification\", \"structured-output\", \"qa-abstention\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"limite-1b-violetto-q80\"], \"profile\": \"smoke\", \"run_group\": \"limite-violetto-q80-smoke\", \"run_id\": \"ea0fd04b-aae3-4020-ac46-99d066560ecb\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"3\"}","model_key":null,"run_id":"ea0fd04b-aae3-4020-ac46-99d066560ecb","sample_id":null,"status":null,"timestamp_utc":"2026-10-20T12:33:29.776734+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"74941c01-91e0-4e9d-bc8e-d883b04c76c5","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"limite-1b-violetto-q80\", \"model_signature\": \"sha256:model:5c5de7987a023b955d7b46893217f72c1c6b1523f722a2097fec543baf5ff73d\", \"provider_key\": \"korgis\", \"run_group\": \"limite-violetto-q80-smoke\", \"run_id\": \"ea0fd04b-aae3-4020-ac46-99d066560ecb\", \"runtime_key\": \"korgis-local\"}","model_key":"limite-1b-violetto-q80","run_id":"ea0fd04b-aae3-4020-ac46-99d066560ecb","sample_id":null,"status":"started","timestamp_utc":"2026-10-20T12:33:29.806387+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":"HTTP transport error: [Errno 61] Connection refused","error_type":"TransportError","event_id":"2fec8ce0-eca0-4638-a3d6-b76397f59c61","event_type":"model.prepare.failed","metadata_json":"{\"model_key\": \"limite-1b-violetto-q80\", \"model_signature\": \"sha256:model:5c5de7987a023b955d7b46893217f72c1c6b1523f722a2097fec543baf5ff73d\", \"provider_key\": \"korgis\", \"run_group\": \"limite-violetto-q80-smoke\", \"run_id\": \"ea0fd04b-aae3-4020-ac46-99d066560ecb\", \"runtime_key\": \"korgis-local\"}","model_key":"limite-1b-violetto-q80","run_id":"ea0fd04b-aae3-4020-ac46-99d066560ecb","sample_id":null,"status":"failed","timestamp_utc":"2026-10-20T12:33:29.824779+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"fa920daa-e449-4403-a604-2dc50ffd314c","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 0, \"completed_cases\": 0, \"failed_cases\": 0, \"model_failures\": 1, \"planned_cases\": 0, \"run_group\": \"limite-violetto-q80-smoke\", \"run_id\": \"ea0fd04b-aae3-4020-ac46-99d066560ecb\", \"skipped_cases\": 0}","model_key":null,"run_id":"ea0fd04b-aae3-4020-ac46-99d066560ecb","sample_id":null,"status":null,"timestamp_utc":"2026-10-20T12:33:29.825206+00:00"}]},"f56a81bc-d911-4b76-9b68-361b14e92a2c":{"cells":[{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T15:34:31.093143+00:00","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":0,"model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.675,"profile":"core","run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-15T15:34:31.093143+00:00","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":0,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.7,"profile":"core","run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"gpt-5.6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","family":"gpt-5.6","model_id":"gpt-5.6-luna","model_key":"gpt-5.6-luna","model_signature":"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8","parameters_b":null,"provider_key":"openai","quantization":null,"run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","runtime_key":"openai-api"},{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"gpt-6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","family":"gpt-6","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","parameters_b":null,"provider_key":"openai","quantization":null,"run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","runtime_key":"openai-api"}],"resources":[],"run":{"completed_at_utc":"2026-10-15T15:34:31.093143+00:00","created_at_utc":"2026-10-15T15:32:09.895446+00:00","git_commit":"1f7d5512e5672ee7d7f22df6958f8e1a9ea03f0c","integrity_state":"VALID","manifest_checksum":"f62609b0fa757f0a535e19e2b0c1b4512904fbe89261fba4bda6dbe5fcbeb173","profile":"core","run_group":"math-v3-core-standalone","run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"3"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6fe30b75-c52f-444f-83a7-ad28db602dc4","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"gpt-5.6-luna\", \"gpt-6-luna\"], \"profile\": \"core\", \"run_group\": \"math-v3-core-standalone\", \"run_id\": \"f56a81bc-d911-4b76-9b68-361b14e92a2c\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"3\"}","model_key":null,"run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","sample_id":null,"status":null,"timestamp_utc":"2026-10-15T15:32:09.895446+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b3274107-517f-4fcf-8b4c-3869be5cf1bd","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"math-v3-core-standalone\", \"run_id\": \"f56a81bc-d911-4b76-9b68-361b14e92a2c\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T15:32:09.912740+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":27.711333008483052,"error_message":null,"error_type":null,"event_id":"4688a89a-a701-4f7b-83e8-55bf5134defd","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"math-v3-core-standalone\", \"run_id\": \"f56a81bc-d911-4b76-9b68-361b14e92a2c\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T15:32:09.940881+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"9b5e7b5e-ac8a-424c-a436-8b956638bef7","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"math-v3-core-standalone\", \"run_id\": \"f56a81bc-d911-4b76-9b68-361b14e92a2c\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T15:32:09.941113+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"63770b7e-51f7-4fb2-8400-ab1872c8a772","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"openai\", \"run_group\": \"math-v3-core-standalone\", \"run_id\": \"f56a81bc-d911-4b76-9b68-361b14e92a2c\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T15:32:09.959082+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"cd236e47-e38a-4d21-ac6f-58d5bdf67703","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"profile\": \"core\", \"provider_key\": \"openai\", \"run_group\": \"math-v3-core-standalone\", \"run_id\": \"f56a81bc-d911-4b76-9b68-361b14e92a2c\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T15:33:18.622032+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3e081c80-a5c8-4f17-866a-21e1b80b6f38","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"math-v3-core-standalone\", \"run_id\": \"f56a81bc-d911-4b76-9b68-361b14e92a2c\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T15:33:18.622410+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.01237494871020317,"error_message":null,"error_type":null,"event_id":"7db21664-0690-4070-9629-7e5af0ef89af","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"provider_key\": \"openai\", \"run_group\": \"math-v3-core-standalone\", \"run_id\": \"f56a81bc-d911-4b76-9b68-361b14e92a2c\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-5.6-luna","run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T15:33:18.622686+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7a7a910b-5382-47ec-8413-b692531dcfe8","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"math-v3-core-standalone\", \"run_id\": \"f56a81bc-d911-4b76-9b68-361b14e92a2c\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T15:33:18.623013+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":1.4382919762283564,"error_message":null,"error_type":null,"event_id":"5514de1d-2446-4f3b-8cc1-786f9e887c6a","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"math-v3-core-standalone\", \"run_id\": \"f56a81bc-d911-4b76-9b68-361b14e92a2c\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T15:33:18.624787+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6736fce6-c612-40a0-bcb7-14643933e6a6","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"math-v3-core-standalone\", \"run_id\": \"f56a81bc-d911-4b76-9b68-361b14e92a2c\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T15:33:18.625103+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"680b8302-b4e2-45d4-98b1-02d04b342aba","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"openai\", \"run_group\": \"math-v3-core-standalone\", \"run_id\": \"f56a81bc-d911-4b76-9b68-361b14e92a2c\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T15:33:18.625626+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c045d721-3acc-4d87-8596-f2503f5e97b7","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"profile\": \"core\", \"provider_key\": \"openai\", \"run_group\": \"math-v3-core-standalone\", \"run_id\": \"f56a81bc-d911-4b76-9b68-361b14e92a2c\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T15:34:31.074619+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"4a704585-38fe-4cbd-891e-22780752159d","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"math-v3-core-standalone\", \"run_id\": \"f56a81bc-d911-4b76-9b68-361b14e92a2c\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","sample_id":null,"status":"started","timestamp_utc":"2026-10-15T15:34:31.075018+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.024916953407227993,"error_message":null,"error_type":null,"event_id":"fcb26e07-0c38-490f-aa43-8fa8806e6a65","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"math-v3-core-standalone\", \"run_id\": \"f56a81bc-d911-4b76-9b68-361b14e92a2c\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T15:34:31.075382+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"26ecaeda-74ae-4c38-94dd-0cf5bb36707a","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-5.6-luna\", \"model_signature\": \"sha256:model:b6168aad56ae75ae50ed93ab57ddcf178358c4eb3b6cb03b33706a99bd8ed3d8\", \"run_group\": \"math-v3-core-standalone\", \"run_id\": \"f56a81bc-d911-4b76-9b68-361b14e92a2c\"}","model_key":"gpt-5.6-luna","run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T15:34:31.090316+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7c3a58be-b753-4089-9e66-b5f63f56e0dc","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:3a5df842430c7a271a8bd20fdeac2dacbf00076135a320efc615fbde8160fbb2\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"run_group\": \"math-v3-core-standalone\", \"run_id\": \"f56a81bc-d911-4b76-9b68-361b14e92a2c\"}","model_key":"gpt-6-luna","run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","sample_id":null,"status":"completed","timestamp_utc":"2026-10-15T15:34:31.092609+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1dced0f0-969e-4b2a-a924-188baac15ef4","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 12, \"completed_cases\": 80, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 80, \"run_group\": \"math-v3-core-standalone\", \"run_id\": \"f56a81bc-d911-4b76-9b68-361b14e92a2c\", \"skipped_cases\": 0}","model_key":null,"run_id":"f56a81bc-d911-4b76-9b68-361b14e92a2c","sample_id":null,"status":null,"timestamp_utc":"2026-10-15T15:34:31.093143+00:00"}]},"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core":{"cells":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T13:59:58.574010+00:00","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":0,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.0,"profile":"core","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"gemini-3.5-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","family":"gemini-3.5","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","parameters_b":null,"provider_key":"gemini","quantization":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core","runtime_key":"gemini-api"}],"resources":[],"run":{"completed_at_utc":"2026-10-13T13:59:58.574010+00:00","created_at_utc":"2026-10-13T13:59:36.980696+00:00","git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","integrity_state":"VALID","manifest_checksum":"65a94b0c058db795af05cc9027c2f034cf3e99dc43057d8fd1a4a1075c34342e","profile":"core","run_group":"gemini-3.5-flash-core-gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a43f740e-7f9e-4ba9-a296-67c0dd2cace9","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"intent-classification\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"gemini-3.5-flash\"], \"profile\": \"core\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:59:36.980696+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ff35344a-7c14-4e22-a8b7-6549ecbd2fec","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:59:37.004094+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":26.503458968363702,"error_message":null,"error_type":null,"event_id":"26b9d648-d46a-49bc-a2b6-2bf1b1243d85","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:59:37.031118+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ad835b9a-d645-4224-8cb3-db1eecc2cf0a","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:59:37.031500+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"45f6c4d6-3e22-4051-87c6-d244743a3b6e","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:59:37.038179+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"40e4ff49-52c3-4344-a660-e51d1370fdbd","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"profile\": \"core\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:59:58.557202+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c8fdb26c-9942-4044-b7d8-2abdcc98942d","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:59:58.557893+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.01654098741710186,"error_message":null,"error_type":null,"event_id":"a72bce6f-1380-4acb-99bf-80ed7bb5aee9","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:59:58.558470+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"9213c271-e850-47b5-bd0a-519a10b45c33","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:59:58.573815+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d397d327-30e6-4fe2-878d-35e2f4d2d8ae","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:59:58.574010+00:00"}]},"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core":{"cells":[{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T13:59:36.375684+00:00","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":0,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.0,"profile":"core","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"gemini-3.5-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","family":"gemini-3.5","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","parameters_b":null,"provider_key":"gemini","quantization":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core","runtime_key":"gemini-api"}],"resources":[],"run":{"completed_at_utc":"2026-10-13T13:59:36.375684+00:00","created_at_utc":"2026-10-13T13:59:30.576539+00:00","git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","integrity_state":"VALID","manifest_checksum":"1f406c4e0a80780e2dc2459c75b6594d2b87c1638e1af8ce6ba6908a46475d3a","profile":"core","run_group":"gemini-3.5-flash-core-gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"367e8e6a-2ca4-4956-967e-f722542d8907","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"gemini-3.5-flash\"], \"profile\": \"core\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:59:30.576539+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d4c9dfff-0bc0-42e6-bd27-4cc80f04e44d","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:59:30.601446+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":40.727583051193506,"error_message":null,"error_type":null,"event_id":"d6e3e60c-ef34-4ef5-9c72-0351e5eea478","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:59:30.642696+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f51d0f90-bcb7-4115-97d4-f31d0d9513cc","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:59:30.643046+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"dae8b89f-fdb5-4de1-9ea1-923c23b7bd1d","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:59:30.655035+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"36a47f32-c1f5-4f7a-9753-0d18d36e07df","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"profile\": \"core\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:59:36.366056+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"af118326-ece0-42b0-8da0-00eea296e38e","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:59:36.367012+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.019167026039212942,"error_message":null,"error_type":null,"event_id":"a937ea2d-dbb6-484b-a29e-1fb28db17853","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:59:36.367514+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2f1d1fa6-6a20-424d-ba00-a2de2c0daa81","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:59:36.375431+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"77bb5176-b602-4032-83ab-db358c418cf8","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 40, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 40, \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:59:36.375684+00:00"}]},"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core":{"cells":[{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T14:00:21.676846+00:00","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":0,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.5,"profile":"core","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"gemini-3.5-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","family":"gemini-3.5","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","parameters_b":null,"provider_key":"gemini","quantization":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core","runtime_key":"gemini-api"}],"resources":[],"run":{"completed_at_utc":"2026-10-13T14:00:21.676846+00:00","created_at_utc":"2026-10-13T13:59:59.227930+00:00","git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","integrity_state":"VALID","manifest_checksum":"4332f4042b677c6ed14aecb1c1f3019cb24ac3e2960ba4477db688f5ce427132","profile":"core","run_group":"gemini-3.5-flash-core-gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7ac9ea9b-2e0f-4cf0-b6df-76246c80a9bd","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"oos-calibration\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"gemini-3.5-flash\"], \"profile\": \"core\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:59:59.227930+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7c893fc5-0ecb-4bd0-becd-467334de05e8","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:59:59.255107+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":29.564417025540024,"error_message":null,"error_type":null,"event_id":"f9e5cf73-a79f-48f4-bfd3-0d3bfc242d34","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:59:59.285246+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0124e8cd-b4cc-4925-9c0d-c71bd20c1642","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:59:59.285543+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"74e0753e-40e2-4a01-94fc-905fc55e8eaf","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:59:59.307966+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f0b73f98-3f73-4c42-9d7c-a65f0a9fabfd","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"profile\": \"core\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T14:00:21.660043+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"55025eb0-6ff0-420d-9d5a-db7a1f6e7158","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T14:00:21.660637+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.012249976862221956,"error_message":null,"error_type":null,"event_id":"7e285f99-23b7-4a8c-a7fc-547556993f9b","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T14:00:21.661448+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"57387503-bc79-43ac-83cb-1b3c758d764a","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T14:00:21.676678+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"895107aa-75aa-4a4f-b300-ab109d5801cc","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T14:00:21.676846+00:00"}]},"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core":{"cells":[{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T13:59:29.849450+00:00","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":0,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.016666666666666666,"profile":"core","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"gemini-3.5-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","family":"gemini-3.5","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","parameters_b":null,"provider_key":"gemini","quantization":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core","runtime_key":"gemini-api"}],"resources":[],"run":{"completed_at_utc":"2026-10-13T13:59:29.849450+00:00","created_at_utc":"2026-10-13T13:59:14.619742+00:00","git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","integrity_state":"VALID","manifest_checksum":"2de33eb1a6cbce9cbf50eda5079efb91adda6fc8e348e5ed0ef860f299a9964e","profile":"core","run_group":"gemini-3.5-flash-core-gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"476e5232-0820-4391-8cc1-a49fbf86275f","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"qa-abstention\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"gemini-3.5-flash\"], \"profile\": \"core\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:59:14.619742+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"04a66ac9-7be1-4f84-9907-090d8e6c0cf6","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:59:14.643831+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":42.23754198756069,"error_message":null,"error_type":null,"event_id":"5e3a080b-696e-4c5c-a3b1-2c2ce4963b3f","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:59:14.686626+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ca05002a-4c73-43e5-8711-1d73ad0e003e","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:59:14.686977+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"713c9adf-f2f8-4b20-b2f9-1cc76e3f1dce","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:59:14.710122+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6615ee2a-7401-4a5d-982d-399c9adc2ffc","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"profile\": \"core\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:59:29.838858+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"202792f1-ce87-4611-9473-271aaee46a69","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:59:29.839236+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.014541961718350649,"error_message":null,"error_type":null,"event_id":"3e4bd1ac-0810-4356-9491-008e49c7391f","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:59:29.839617+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"14513086-f05a-44fd-b4f3-b36bb607dcc1","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:59:29.849213+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"fafd45c8-5403-4c6e-9c7a-c4526cee5e38","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:59:29.849450+00:00"}]},"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core":{"cells":[{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T13:59:13.872786+00:00","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":0,"model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.03333333333333333,"profile":"core","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"gemini-3.5-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","family":"gemini-3.5","model_id":"gemini-3.5-flash","model_key":"gemini-3.5-flash","model_signature":"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a","parameters_b":null,"provider_key":"gemini","quantization":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core","runtime_key":"gemini-api"}],"resources":[],"run":{"completed_at_utc":"2026-10-13T13:59:13.872786+00:00","created_at_utc":"2026-10-13T13:57:42.057717+00:00","git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","integrity_state":"VALID","manifest_checksum":"a610f8b5dfa4e76b8a6043d9edc233307df20864150318d34809d985e17ebd60","profile":"core","run_group":"gemini-3.5-flash-core-gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6e132123-19af-4eeb-83ce-c8b1bc61610e","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"structured-output\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"gemini-3.5-flash\"], \"profile\": \"core\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:57:42.057717+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6549bb6b-0cce-43a9-a4ef-e2795d82053e","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:57:42.079755+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":43.41550002573058,"error_message":null,"error_type":null,"event_id":"702ef725-b5c3-4e05-9f27-a59ef410f527","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:57:42.123681+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"db6542ff-5bb0-4358-9ea5-588279bac7ce","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:57:42.124014+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0dbeacb2-717e-45ad-8b62-ad5dd4a93aae","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:57:42.175338+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"700d8e8e-aa0b-4bec-a227-1cc55b8e5442","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"profile\": \"core\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:59:13.861892+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"819557aa-c563-475f-aa83-952e4ccba413","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:59:13.862496+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.015666009858250618,"error_message":null,"error_type":null,"event_id":"87e3897a-6be4-4389-84e8-53b1a139460a","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:59:13.863013+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e604d8d8-6162-4f44-9dcd-0baf08dd784f","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.5-flash\", \"model_signature\": \"sha256:model:16c9f32705232e1f4462c937b8c81c949197675ccd0234b08ca18f633912ac6a\", \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core\"}","model_key":"gemini-3.5-flash","run_id":"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:59:13.872643+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e0f80e4f-147d-4467-897e-ccc40b334a2f","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"gemini-3.5-flash-core-gemini-3.5-flash\", \"run_id\": \"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"gemini-3.5-flash-core-gemini-3.5-flash-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:59:13.872786+00:00"}]},"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core":{"cells":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T13:39:20.878424+00:00","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":0,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.0,"profile":"core","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"gemini-3.8-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","family":"gemini-3.8","model_id":"gemini-3.8-flash","model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","parameters_b":null,"provider_key":"gemini","quantization":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core","runtime_key":"gemini-api"}],"resources":[],"run":{"completed_at_utc":"2026-10-13T13:39:20.878424+00:00","created_at_utc":"2026-10-13T13:38:59.179423+00:00","git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","integrity_state":"VALID","manifest_checksum":"66320c97b26907d24749e7e5a4bc573f9aea3a77c6b57591f67c844b8c799165","profile":"core","run_group":"gemini-3.8-flash-core-gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"89efb01b-c016-4dfb-a557-de60668c363b","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"intent-classification\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"gemini-3.8-flash\"], \"profile\": \"core\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:38:59.179423+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0127581e-2f4f-4f67-83ff-3f756206b63a","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:38:59.201113+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":22.093582956586033,"error_message":null,"error_type":null,"event_id":"470ac609-f457-44a3-82a4-c4b4864fb936","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:38:59.223604+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"23a4cf69-77a6-4eb2-a9cc-8d66c0c966e0","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:38:59.223868+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e35f0ea7-aba7-44f7-8151-532a9101fb4b","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:38:59.229415+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"9cda15c0-c2fe-4969-a4f0-982fdec2306b","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"profile\": \"core\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:39:20.854970+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"de685908-76ac-4ef7-a691-921ccf4cc723","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:39:20.855707+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.0415419926866889,"error_message":null,"error_type":null,"event_id":"9b4367d9-2ef8-4b73-b53e-1e2d0c3cb29a","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:39:20.856339+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0d634786-ffc0-478a-b1a2-1a45b2266b58","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:39:20.878098+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"23c85c43-6068-4176-bd60-6bf51f24b6c7","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:39:20.878424+00:00"}]},"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core":{"cells":[{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T13:38:58.640898+00:00","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":0,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.0,"profile":"core","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"gemini-3.8-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","family":"gemini-3.8","model_id":"gemini-3.8-flash","model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","parameters_b":null,"provider_key":"gemini","quantization":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core","runtime_key":"gemini-api"}],"resources":[],"run":{"completed_at_utc":"2026-10-13T13:38:58.640898+00:00","created_at_utc":"2026-10-13T13:38:52.779684+00:00","git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","integrity_state":"VALID","manifest_checksum":"e515f3d8c9031f07bdf64c3c31320a05b8d6f8fdd0493427e591d611ad9702f6","profile":"core","run_group":"gemini-3.8-flash-core-gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"25bef6ec-91b8-4d8e-88ce-fc2f8ddb8826","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"gemini-3.8-flash\"], \"profile\": \"core\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:38:52.779684+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"fd1b86b4-31b6-4914-ba56-695ad8e7aaa7","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:38:52.797118+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":31.81908297119662,"error_message":null,"error_type":null,"event_id":"a5ddc5ab-5950-46f5-90b9-997c9ed4e9d8","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:38:52.829351+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8179a557-ee1b-40e0-bd47-496dabd62efa","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:38:52.829606+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"fc7408ed-6961-44f9-b7c7-58d3e2317f15","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:38:52.840953+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"01a64c13-c7af-4848-9726-0a52c42829fd","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"profile\": \"core\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:38:58.632341+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a9b73f1a-28d2-4db2-92d6-2db2c6477ea0","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:38:58.632833+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.025416025891900063,"error_message":null,"error_type":null,"event_id":"a770da12-5e76-418b-9a1b-09585fe2729f","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:38:58.633348+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"17f5c9a9-4036-4a5b-84f2-5da3f34b9372","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:38:58.640421+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"db16b66a-9e35-4f42-b656-1a32ec3fbe46","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 40, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 40, \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:38:58.640898+00:00"}]},"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core":{"cells":[{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T13:39:43.452981+00:00","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":0,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.5,"profile":"core","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"gemini-3.8-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","family":"gemini-3.8","model_id":"gemini-3.8-flash","model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","parameters_b":null,"provider_key":"gemini","quantization":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core","runtime_key":"gemini-api"}],"resources":[],"run":{"completed_at_utc":"2026-10-13T13:39:43.452981+00:00","created_at_utc":"2026-10-13T13:39:21.674629+00:00","git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","integrity_state":"VALID","manifest_checksum":"e513e0019f8a657fda866fb0a280b5b812750b2fe87706e8435fd6f8743e2df3","profile":"core","run_group":"gemini-3.8-flash-core-gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0c0e1de9-0328-4518-8681-5162afb4947a","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"oos-calibration\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"gemini-3.8-flash\"], \"profile\": \"core\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:39:21.674629+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3285d758-f8a3-4abf-9d1f-14dd1a442977","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:39:21.693167+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":31.89995902357623,"error_message":null,"error_type":null,"event_id":"89d6bf0d-9ce7-49a2-8071-72412be9059d","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:39:21.725473+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"01268471-24e1-461e-a3f2-bb914336c5a1","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:39:21.725685+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"93c6ccc2-1ccf-407b-a2ad-8eefeff7ca02","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:39:21.744312+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"fc628a61-0220-4e45-9dce-863c3291a525","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"profile\": \"core\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:39:43.434439+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8f4b6ce1-ebeb-44f0-b5b2-124f09824b05","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:39:43.435309+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.03850000211969018,"error_message":null,"error_type":null,"event_id":"a30ed0b6-4b59-439d-8111-c324c920025e","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:39:43.436064+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"de19cf41-1164-4216-b3fc-e2c62847c1c7","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:39:43.452706+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3fdc54fe-a6a1-430c-a2b7-6a2617a2040c","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:39:43.452981+00:00"}]},"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core":{"cells":[{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T13:38:52.091017+00:00","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":0,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.0,"profile":"core","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"gemini-3.8-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","family":"gemini-3.8","model_id":"gemini-3.8-flash","model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","parameters_b":null,"provider_key":"gemini","quantization":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core","runtime_key":"gemini-api"}],"resources":[],"run":{"completed_at_utc":"2026-10-13T13:38:52.091017+00:00","created_at_utc":"2026-10-13T13:38:43.199673+00:00","git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","integrity_state":"VALID","manifest_checksum":"7159eeaf81f52e1c378f63018321854dca2dae384b29be3a35ca3f3202f60fad","profile":"core","run_group":"gemini-3.8-flash-core-gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"657227ef-0ac9-49db-9559-7a00d8d9916f","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"qa-abstention\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"gemini-3.8-flash\"], \"profile\": \"core\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:38:43.199673+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a53bd98a-ccd7-4d29-ac2f-360a1b531fbd","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:38:43.221172+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":34.96387501945719,"error_message":null,"error_type":null,"event_id":"b519a53e-fa15-4121-a42c-362fc59f126b","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:38:43.256518+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"9d24a2e4-b58f-4aac-a8dd-def0172259c8","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:38:43.256714+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"30e9239c-3ecd-416d-9736-59aee866fd4b","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:38:43.277318+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"01d4dd83-3472-4321-9a89-d7698de2adac","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"profile\": \"core\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:38:52.078402+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"305a6026-e715-41fb-90a0-1298090dd130","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:38:52.079348+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.038624973967671394,"error_message":null,"error_type":null,"event_id":"e139d175-0771-4549-b391-39586364f5f8","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:38:52.080357+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1fef860a-898a-4d73-b37f-6d18b7a467c4","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:38:52.090619+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ad9f0db1-6810-4a24-b13d-cd8d4c8260ff","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:38:52.091017+00:00"}]},"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core":{"cells":[{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T13:38:42.203597+00:00","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","failure_count":0,"model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.0,"profile":"core","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"gemini-3.8-flash","execution_signature":"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d","family":"gemini-3.8","model_id":"gemini-3.8-flash","model_key":"gemini-3.8-flash","model_signature":"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e","parameters_b":null,"provider_key":"gemini","quantization":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core","runtime_key":"gemini-api"}],"resources":[],"run":{"completed_at_utc":"2026-10-13T13:38:42.203597+00:00","created_at_utc":"2026-10-13T13:33:41.986765+00:00","git_commit":"ca96fac25b6cb39a48db88054ceaa4158417250f","integrity_state":"VALID","manifest_checksum":"b22b72bfc27d6243f87cf7287b680e9e1cbfee60f0bab91eb971183aefe971b5","profile":"core","run_group":"gemini-3.8-flash-core-gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"840bfeea-c29e-4ada-bd7a-df77518ec5b9","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"structured-output\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"gemini-3.8-flash\"], \"profile\": \"core\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:33:41.986765+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"65ad6efa-9fe1-482d-a5f4-b525569bd462","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:33:42.008445+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":35.30650003813207,"error_message":null,"error_type":null,"event_id":"1f4f448f-4b0d-421e-a880-ad0fec416d34","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:33:42.044150+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1fc42c6b-8662-4df2-9155-d52e587b5503","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:33:42.044359+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"4b34c67f-e4d3-4296-bf32-3914e439c647","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:33:42.091603+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"aea66c33-9f38-4db1-a2a7-0f23733487f2","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"profile\": \"core\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:38:42.194266+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e92bd9c8-e0fe-4ac1-8f95-db3e981afb0e","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:38:42.194814+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.02237502485513687,"error_message":null,"error_type":null,"event_id":"e926f4ce-4150-43d5-a348-630abea40964","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"provider_key\": \"gemini\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core\", \"runtime_key\": \"gemini-api\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:38:42.195177+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"54e1a387-c147-4f9a-a8e4-7f76e69a3d05","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:257d602197873d3220812b4440705e8140c8c5667abc36217c12df24452b927d\", \"model_key\": \"gemini-3.8-flash\", \"model_signature\": \"sha256:model:d519edb0c9ee5bc853913442433dbaa52660105ab7e83d96e99ba1164f4e9d4e\", \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core\"}","model_key":"gemini-3.8-flash","run_id":"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:38:42.203308+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c4a3a7c4-ffd3-4029-b4a8-a718de566684","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"gemini-3.8-flash-core-gemini-3.8-flash\", \"run_id\": \"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"gemini-3.8-flash-core-gemini-3.8-flash-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:38:42.203597+00:00"}]},"gpt-6-luna-core-gpt-6-luna-intent-classification-core":{"cells":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T13:18:34.815598+00:00","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":0,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.7987012987012987,"profile":"core","run_id":"gpt-6-luna-core-gpt-6-luna-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"gpt-6-luna-core-gpt-6-luna-intent-classification-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"gpt-6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","family":"gpt-6","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","parameters_b":null,"provider_key":"openai","quantization":null,"run_id":"gpt-6-luna-core-gpt-6-luna-intent-classification-core","runtime_key":"openai-api"}],"resources":[],"run":{"completed_at_utc":"2026-10-13T13:18:34.815598+00:00","created_at_utc":"2026-10-13T13:13:21.560235+00:00","git_commit":"7f796c10d99e934da6417ea54405fbf9352ad399","integrity_state":"VALID","manifest_checksum":"93b885c86aa5dcdb46375a7880e376c4bc80ffa5f0cfc3c8e6cd2b3871e6929a","profile":"core","run_group":"gpt-6-luna-core-gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-intent-classification-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3d8c120b-abf0-4013-a248-f2725d4925eb","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"intent-classification\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"gpt-6-luna\"], \"profile\": \"core\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-intent-classification-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"gpt-6-luna-core-gpt-6-luna-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:13:21.560235+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f1cebccd-9e8c-482f-970e-273eaf2f96dc","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-intent-classification-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:13:21.580707+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":37.045667006168514,"error_message":null,"error_type":null,"event_id":"3a0f5965-78d5-45b3-9b99-3078a1bf8be2","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-intent-classification-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:13:21.618189+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"894a9c7b-0ed5-4a7f-a69c-3d2ceea56fd6","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-intent-classification-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:13:21.618403+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2c41a004-c6fa-436d-82d8-e5c63a83e13b","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-intent-classification-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:13:21.624414+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"91e94b36-9419-45d1-bbf6-0bf3f15425f6","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"profile\": \"core\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-intent-classification-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:18:34.795044+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d82bc98d-84ec-406a-87c5-dfe00b74eaad","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-intent-classification-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:18:34.795516+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.019041995983570814,"error_message":null,"error_type":null,"event_id":"84d9b3a8-834e-4082-baff-035ec9ce5090","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-intent-classification-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:18:34.795993+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8e70154b-943f-4885-9a9f-b8649e2c466e","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-intent-classification-core\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:18:34.815444+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ab9de501-a024-4ddd-a5a9-add7dfb234d4","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-intent-classification-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"gpt-6-luna-core-gpt-6-luna-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:18:34.815598+00:00"}]},"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core":{"cells":[{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T13:13:20.819089+00:00","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":0,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.425,"profile":"core","run_id":"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"gpt-6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","family":"gpt-6","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","parameters_b":null,"provider_key":"openai","quantization":null,"run_id":"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core","runtime_key":"openai-api"}],"resources":[],"run":{"completed_at_utc":"2026-10-13T13:13:20.819089+00:00","created_at_utc":"2026-10-13T13:12:08.510262+00:00","git_commit":"7f796c10d99e934da6417ea54405fbf9352ad399","integrity_state":"VALID","manifest_checksum":"92b1fc96ce25aab116c0c88369039301d326335d870c17ce1b94ddf9d7036505","profile":"core","run_group":"gpt-6-luna-core-gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"30e4e8f4-02b3-4536-877d-7a849b7dd8f9","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"gpt-6-luna\"], \"profile\": \"core\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:12:08.510262+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7de970a3-4e12-4d22-92a0-d09e785ec2e6","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:12:08.533636+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":41.84020799584687,"error_message":null,"error_type":null,"event_id":"516e5591-43ce-4e4e-b589-feaa47d65ce5","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:12:08.575909+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"90ba6504-c869-4ec1-a806-cacc34135867","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:12:08.576139+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"9d009118-1f3d-4880-b08b-735ce81713bb","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:12:08.587430+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2e842c75-f1ea-4f0c-a921-66d67b7994b5","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"profile\": \"core\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:13:20.813214+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"fe3c7eee-9c63-4836-8753-e519b849bf53","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:13:20.813436+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.0158340553753078,"error_message":null,"error_type":null,"event_id":"2d639b1f-2404-4333-b99c-4ef999abf035","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:13:20.813655+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2a31e588-27bd-4f8d-b560-b6c2e1fe19f4","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:13:20.818931+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6898e453-a5fc-4ffa-b14c-de3c129196c9","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 40, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 40, \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"gpt-6-luna-core-gpt-6-luna-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:13:20.819089+00:00"}]},"gpt-6-luna-core-gpt-6-luna-oos-calibration-core":{"cells":[{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T13:25:20.495814+00:00","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":0,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.9935064935064936,"profile":"core","run_id":"gpt-6-luna-core-gpt-6-luna-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"gpt-6-luna-core-gpt-6-luna-oos-calibration-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"gpt-6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","family":"gpt-6","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","parameters_b":null,"provider_key":"openai","quantization":null,"run_id":"gpt-6-luna-core-gpt-6-luna-oos-calibration-core","runtime_key":"openai-api"}],"resources":[],"run":{"completed_at_utc":"2026-10-13T13:25:20.495814+00:00","created_at_utc":"2026-10-13T13:18:35.677977+00:00","git_commit":"7f796c10d99e934da6417ea54405fbf9352ad399","integrity_state":"VALID","manifest_checksum":"8f2065a85a451b584b2820e97c463bc4aeac45ec5a256060a114d2120df77928","profile":"core","run_group":"gpt-6-luna-core-gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-oos-calibration-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1ab7ca5d-0ce8-44ed-9c3a-4b3c273facf6","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"oos-calibration\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"gpt-6-luna\"], \"profile\": \"core\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-oos-calibration-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"gpt-6-luna-core-gpt-6-luna-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:18:35.677977+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ebf22aa0-6fd6-4f7c-b24a-09303f5fedb6","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-oos-calibration-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:18:35.696497+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":52.97554098069668,"error_message":null,"error_type":null,"event_id":"bf7fff89-412b-4a5b-a167-39fed1517482","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-oos-calibration-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:18:35.750300+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"bca08719-58d6-43cd-9778-321c83b62770","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-oos-calibration-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:18:35.750738+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"49be5d6c-3271-49cd-98eb-bb1682c389ed","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-oos-calibration-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:18:35.781086+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"cee72dc9-4cf8-4ff2-93b2-6be43a615d44","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"profile\": \"core\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-oos-calibration-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:25:20.475316+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3d9bc4ba-08c9-4004-ac1a-732876f7dc37","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-oos-calibration-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:25:20.475695+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.030166993383318186,"error_message":null,"error_type":null,"event_id":"e57e12af-60e8-47e9-9030-bf84d75b5de7","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-oos-calibration-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:25:20.476250+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7e898722-b5e5-4cda-a29a-50c19392d24d","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-oos-calibration-core\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:25:20.495445+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0c76bc99-c492-4a1a-a541-5bfa2bb0903b","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-oos-calibration-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"gpt-6-luna-core-gpt-6-luna-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:25:20.495814+00:00"}]},"gpt-6-luna-core-gpt-6-luna-qa-abstention-core":{"cells":[{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T13:12:07.696210+00:00","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":0,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.95,"profile":"core","run_id":"gpt-6-luna-core-gpt-6-luna-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"gpt-6-luna-core-gpt-6-luna-qa-abstention-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"gpt-6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","family":"gpt-6","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","parameters_b":null,"provider_key":"openai","quantization":null,"run_id":"gpt-6-luna-core-gpt-6-luna-qa-abstention-core","runtime_key":"openai-api"}],"resources":[],"run":{"completed_at_utc":"2026-10-13T13:12:07.696210+00:00","created_at_utc":"2026-10-13T13:10:28.239118+00:00","git_commit":"7f796c10d99e934da6417ea54405fbf9352ad399","integrity_state":"VALID","manifest_checksum":"9f484a3eddd6bc684cc69e3e3bb6b0ce561689ca18a7664a7d87fc3e0e56864d","profile":"core","run_group":"gpt-6-luna-core-gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-qa-abstention-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0ce1f948-27d9-463a-a302-46ab25a3d7ff","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"qa-abstention\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"gpt-6-luna\"], \"profile\": \"core\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-qa-abstention-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"gpt-6-luna-core-gpt-6-luna-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:10:28.239118+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"cb7b3f19-0beb-48dd-b68c-c2b919756946","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-qa-abstention-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:10:28.265224+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":88.52849999675527,"error_message":null,"error_type":null,"event_id":"d54c2a38-acba-41cc-870e-4b9a5eb58ed6","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-qa-abstention-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:10:28.355683+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b41396e9-b89d-4039-bc54-9f0dddf79489","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-qa-abstention-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:10:28.356079+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"977f41a4-537f-46ea-9bc0-efd550da529a","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-qa-abstention-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:10:28.382071+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"90a47bf4-5513-40f8-8019-26206e22c199","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"profile\": \"core\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-qa-abstention-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:12:07.683746+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1cb525cd-b7df-4ce1-b7d9-6029ed61219b","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-qa-abstention-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:12:07.684378+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.017624988686293364,"error_message":null,"error_type":null,"event_id":"3af6842d-7b52-4f3b-a85c-0b260c3fa25c","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-qa-abstention-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:12:07.684942+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2ad4097e-e5ef-4be2-94dc-c1f4d3f2b553","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-qa-abstention-core\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:12:07.695689+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"468d611a-102d-4b05-a6c6-ea4a92c21a5c","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-qa-abstention-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"gpt-6-luna-core-gpt-6-luna-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:12:07.696210+00:00"}]},"gpt-6-luna-core-gpt-6-luna-structured-output-core":{"cells":[{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T13:10:27.524208+00:00","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","failure_count":0,"model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.8333333333333334,"profile":"core","run_id":"gpt-6-luna-core-gpt-6-luna-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"gpt-6-luna-core-gpt-6-luna-structured-output-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":null,"artifact_size_bytes":null,"deployment":"api","effective_model_id":"gpt-6-luna","execution_signature":"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce","family":"gpt-6","model_id":"gpt-6-luna","model_key":"gpt-6-luna","model_signature":"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a","parameters_b":null,"provider_key":"openai","quantization":null,"run_id":"gpt-6-luna-core-gpt-6-luna-structured-output-core","runtime_key":"openai-api"}],"resources":[],"run":{"completed_at_utc":"2026-10-13T13:10:27.524208+00:00","created_at_utc":"2026-10-13T13:08:11.184912+00:00","git_commit":"7f796c10d99e934da6417ea54405fbf9352ad399","integrity_state":"VALID","manifest_checksum":"6c78148d0e40f2c11c638c0136561f23d20563d8da87816453b2cb9005c702fc","profile":"core","run_group":"gpt-6-luna-core-gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-structured-output-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"250e7eb4-4dd5-4a5a-a342-734151d331e3","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"structured-output\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"gpt-6-luna\"], \"profile\": \"core\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-structured-output-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"gpt-6-luna-core-gpt-6-luna-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:08:11.184912+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"fa547adf-0196-4b19-936e-5874f88bfa81","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-structured-output-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:08:11.207623+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":37.8475830075331,"error_message":null,"error_type":null,"event_id":"79b55dee-749e-4889-812c-2431560bcad3","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-structured-output-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:08:11.245968+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b4efb420-969c-48e8-8fa3-206bb75d1dff","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-structured-output-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:08:11.246246+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"45c6181d-6644-4fb3-a3d5-d2f8688882dd","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-structured-output-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:08:11.297650+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"864256da-8eca-43f5-ab5b-eb0dce7e0635","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"profile\": \"core\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-structured-output-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:10:27.511076+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"46f6320e-4f78-4913-a36e-630344c3e2d3","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-structured-output-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T13:10:27.511503+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":0.02441596006974578,"error_message":null,"error_type":null,"event_id":"afdbb7c6-0c94-4248-ad19-15ba0d5dc37c","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"provider_key\": \"openai\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-structured-output-core\", \"runtime_key\": \"openai-api\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:10:27.511904+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"5020257d-ce38-4a94-a2a2-22582d93f030","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:0844c7e55e08cd501798151c8e2a3bfe8f24250135ab2d0b17374e7ee79bf4ce\", \"model_key\": \"gpt-6-luna\", \"model_signature\": \"sha256:model:0b1bd70788f66e54dec061e0d3c304e689c40d9105b2731e3889de763cf44d1a\", \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-structured-output-core\"}","model_key":"gpt-6-luna","run_id":"gpt-6-luna-core-gpt-6-luna-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T13:10:27.523959+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a7316b9f-a4d9-4eb0-8ba2-1083f316fe37","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"gpt-6-luna-core-gpt-6-luna\", \"run_id\": \"gpt-6-luna-core-gpt-6-luna-structured-output-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"gpt-6-luna-core-gpt-6-luna-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T13:10:27.524208+00:00"}]},"lfm25-consolidated-latest":{"cells":[{"benchmark_signature":"sha256:consolidated:42ab94943101512840504e431240a2dc1e9c2c51d6e055498c7b6b44c803f515","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-22T13:12:52.242438+00:00","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.7012987012987013,"profile":"core","run_id":"lfm25-consolidated-latest","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"},{"benchmark_signature":"sha256:consolidated:20f65d6394555078bc9dce0c1393e10d963aecb84ce9344ece12046a78d42de3","capability_id":"mathematical-reasoning","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-22T13:12:52.242438+00:00","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.95,"profile":"core","run_id":"lfm25-consolidated-latest","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"},{"benchmark_signature":"sha256:consolidated:9468d75aebe9e805cefc3643f61a9e5fe6127c7a4e3a987608b30841b9bc9fa0","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-22T13:12:52.242438+00:00","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.9545454545454546,"profile":"core","run_id":"lfm25-consolidated-latest","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"},{"benchmark_signature":"sha256:consolidated:b832fe612fcfe138845fe193a557c686f68ecb2f59d1b3f4784e653e9ec82e5c","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-22T13:12:52.242438+00:00","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.9666666666666667,"profile":"core","run_id":"lfm25-consolidated-latest","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"},{"benchmark_signature":"sha256:consolidated:fd230e4e9f030196ab13d72a0beb260f92fbe3bfb1cfa331ed8e0678b55123f9","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-22T13:12:52.242438+00:00","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7666666666666667,"profile":"core","run_id":"lfm25-consolidated-latest","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"configuration":{"changed_dimension":null,"configuration_id":"consolidated","experiment_kind":"consolidated","inference_json":"{\"generation_configs\": [{\"extra\": {}, \"max_output_tokens\": 2048, \"seed\": 42, \"stop\": [], \"temperature\": 0.0}, {\"extra\": {}, \"max_output_tokens\": 4096, \"seed\": 42, \"stop\": [], \"temperature\": 0.0}], \"mixed_generation\": true}","is_baseline":false,"label":null,"run_id":"lfm25-consolidated-latest","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":10947100928,"deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_signature":"sha256:execution:483084303d22fa6970f40d15a426610a84ae33e25402c7a795fe05db7bcbd262","family":"lfm2.5","model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"provider_key":"korgis","quantization":"UD-Q3_K_XL","run_id":"lfm25-consolidated-latest","runtime_key":"korgis-local"}],"resources":[],"run":{"completed_at_utc":"2026-10-22T13:12:52.242438+00:00","created_at_utc":"2026-10-22T13:12:52.242438+00:00","git_commit":null,"integrity_state":"VALID","manifest_checksum":"2c2c5c7f7d18b2a1f0abf8ca525473967017f29964aac087ec0b38c00b6d1719","profile":"core","run_group":"lfm25-consolidated","run_id":"lfm25-consolidated-latest","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"3"},"schema_version":"1","timeline":[]},"lfm25-math-v2-core":{"cells":[{"benchmark_signature":"sha256:benchmark:600e8b1d94c1e2bc55bd8802918ce0c6f18dece6d8a9c02247a530fc7b015409","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-22T14:07:00.510877+00:00","execution_signature":"sha256:execution:503d302009e21bb18588b87ac3897f9112994a479905bbca872384efa4360298","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.775,"profile":"core","run_id":"lfm25-math-v2-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{\"max_output_tokens\": 4096}","is_baseline":false,"label":null,"run_id":"lfm25-math-v2-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":10947100928,"deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_signature":"sha256:execution:503d302009e21bb18588b87ac3897f9112994a479905bbca872384efa4360298","family":"lfm2.5","model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"provider_key":"korgis","quantization":"UD-Q3_K_XL","run_id":"lfm25-math-v2-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:503d302009e21bb18588b87ac3897f9112994a479905bbca872384efa4360298","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","process_cpu_percent_avg":10.29314878109045,"process_rss_bytes_avg":1417688467.7424607,"process_rss_bytes_peak":1463648256.0,"sample_count":1316,"sample_interval_ms":313.23082923487743,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4090920960.0}],"run":{"completed_at_utc":"2026-10-22T14:07:00.510877+00:00","created_at_utc":"2026-10-22T13:59:54.129141+00:00","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","integrity_state":"VALID","manifest_checksum":"f088d8eeac7c8f9f978a001a29290a52b1e28f6143178cfb49cea1b6eeac8da3","profile":"core","run_group":"lfm25-math-v2","run_id":"lfm25-math-v2-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"3"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"71652812-362f-4df9-a4e4-47450de16bac","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning-v2\"], \"configuration_id\": \"default\", \"inference_config\": {\"max_output_tokens\": 4096}, \"model_keys\": [\"lfm2.5-24b-a2b-ud-q3-k-xl\"], \"profile\": \"core\", \"run_group\": \"lfm25-math-v2\", \"run_id\": \"lfm25-math-v2-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"3\"}","model_key":null,"run_id":"lfm25-math-v2-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-22T13:59:54.129141+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8879721d-2246-47d2-89a5-c66a4d2c6017","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"lfm25-math-v2\", \"run_id\": \"lfm25-math-v2-core\", \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-math-v2-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T13:59:54.152218+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":21249.07645804342,"error_message":null,"error_type":null,"event_id":"54759ae5-1ce6-486c-8d2c-0076726689cd","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:503d302009e21bb18588b87ac3897f9112994a479905bbca872384efa4360298\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"lfm25-math-v2\", \"run_id\": \"lfm25-math-v2-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-math-v2-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T14:00:15.401962+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8acd410d-bce9-49dd-8b65-2e1d3df072c7","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning-v2\", \"execution_signature\": \"sha256:execution:503d302009e21bb18588b87ac3897f9112994a479905bbca872384efa4360298\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"lfm25-math-v2\", \"run_id\": \"lfm25-math-v2-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-math-v2-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T14:00:15.403089+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b63f7574-42df-4867-8bb8-d72241f9887a","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:600e8b1d94c1e2bc55bd8802918ce0c6f18dece6d8a9c02247a530fc7b015409\", \"capability_id\": \"mathematical-reasoning-v2\", \"execution_signature\": \"sha256:execution:503d302009e21bb18588b87ac3897f9112994a479905bbca872384efa4360298\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"lfm25-math-v2\", \"run_id\": \"lfm25-math-v2-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-math-v2-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T14:00:15.414995+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"4b8a4f70-7f7c-4523-9310-fd7d386a132d","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:600e8b1d94c1e2bc55bd8802918ce0c6f18dece6d8a9c02247a530fc7b015409\", \"capability_id\": \"mathematical-reasoning-v2\", \"execution_signature\": \"sha256:execution:503d302009e21bb18588b87ac3897f9112994a479905bbca872384efa4360298\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"lfm25-math-v2\", \"run_id\": \"lfm25-math-v2-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-math-v2-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T14:07:00.259655+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"18c627e3-dfa8-4b77-8812-36fa171ba378","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:503d302009e21bb18588b87ac3897f9112994a479905bbca872384efa4360298\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"lfm25-math-v2\", \"run_id\": \"lfm25-math-v2-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-math-v2-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T14:07:00.259880+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":244.13579201791435,"error_message":null,"error_type":null,"event_id":"415c7198-9064-4bec-bbce-756ecc744764","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:503d302009e21bb18588b87ac3897f9112994a479905bbca872384efa4360298\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"lfm25-math-v2\", \"run_id\": \"lfm25-math-v2-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-math-v2-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T14:07:00.504246+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d643a167-0d17-4227-b721-4961ce7ef910","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:600e8b1d94c1e2bc55bd8802918ce0c6f18dece6d8a9c02247a530fc7b015409\", \"capability_id\": \"mathematical-reasoning-v2\", \"execution_signature\": \"sha256:execution:503d302009e21bb18588b87ac3897f9112994a479905bbca872384efa4360298\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"run_group\": \"lfm25-math-v2\", \"run_id\": \"lfm25-math-v2-core\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-math-v2-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T14:07:00.510677+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a51a4351-e3d9-4991-b3b2-cf5b2ef42124","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 40, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 40, \"run_group\": \"lfm25-math-v2\", \"run_id\": \"lfm25-math-v2-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"lfm25-math-v2-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-22T14:07:00.510877+00:00"}]},"lfm25-recovery-4096":{"cells":[{"benchmark_signature":"sha256:benchmark:3e83e05786bfa67a0e16cb28b57ec52fff1967e5e59a7bb032185b9df349aa2d","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-22T12:49:48.793583+00:00","execution_signature":"sha256:execution:641822ff7e4fc678a816a765bd43258a89b949aad2eec0a36c6fb22e990053ee","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.2857142857142857,"profile":"core","run_id":"lfm25-recovery-4096","sample_count":28,"status":"COMPLETED","task_id":"intent-classification"},{"benchmark_signature":"sha256:benchmark:b7c8f155e579d1d1d981fb2d54369194d1a8415fe5833ed759f937156c77e621","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-22T12:49:48.793583+00:00","execution_signature":"sha256:execution:641822ff7e4fc678a816a765bd43258a89b949aad2eec0a36c6fb22e990053ee","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.8,"profile":"core","run_id":"lfm25-recovery-4096","sample_count":20,"status":"COMPLETED","task_id":"calibrated-intent-classification"},{"benchmark_signature":"sha256:benchmark:c26dc80a4a3d7172c20aaf4a0d946e31de5730fcbc2b0e3ef1656e2f69c25727","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-22T12:49:48.793583+00:00","execution_signature":"sha256:execution:641822ff7e4fc678a816a765bd43258a89b949aad2eec0a36c6fb22e990053ee","failure_count":0,"model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.0,"profile":"core","run_id":"lfm25-recovery-4096","sample_count":1,"status":"COMPLETED","task_id":"structured-output"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"recovery","inference_json":"{\"max_output_tokens\": 4096}","is_baseline":false,"label":null,"run_id":"lfm25-recovery-4096","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":10947100928,"deployment":"local","effective_model_id":"lfm2.5-24b-a2b-ud-q3-k-xl","execution_signature":"sha256:execution:641822ff7e4fc678a816a765bd43258a89b949aad2eec0a36c6fb22e990053ee","family":"lfm2.5","model_id":"unsloth/LFM2.5-24B-A2B-GGUF","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","model_signature":"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce","parameters_b":24.0,"provider_key":"korgis","quantization":"UD-Q3_K_XL","run_id":"lfm25-recovery-4096","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:641822ff7e4fc678a816a765bd43258a89b949aad2eec0a36c6fb22e990053ee","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","process_cpu_percent_avg":10.013862554128162,"process_rss_bytes_avg":6110886262.190096,"process_rss_bytes_peak":6236717056.0,"sample_count":4904,"sample_interval_ms":321.4307997644429,"sampling_error_count":2,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":3356114944.0},{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:641822ff7e4fc678a816a765bd43258a89b949aad2eec0a36c6fb22e990053ee","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","process_cpu_percent_avg":10.803087480500368,"process_rss_bytes_avg":6089185261.083659,"process_rss_bytes_peak":6233079808.0,"sample_count":3920,"sample_interval_ms":323.1321839458206,"sampling_error_count":1,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4036706304.0},{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:641822ff7e4fc678a816a765bd43258a89b949aad2eec0a36c6fb22e990053ee","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","process_cpu_percent_avg":11.98296171896892,"process_rss_bytes_avg":6063619627.885715,"process_rss_bytes_peak":6169853952.0,"sample_count":210,"sample_interval_ms":321.0299792632674,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4441014272.0}],"run":{"completed_at_utc":"2026-10-22T12:49:48.793583+00:00","created_at_utc":"2026-10-22T12:01:07.926009+00:00","git_commit":"12ee8e19c04f9f3d44a37e0086a87209d7fc2ce9","integrity_state":"VALID","manifest_checksum":"146cea2f3b0118bec06dfff67ba54a9492d6c33e6a530a444d97be1acd6f00e6","profile":"core","run_group":"lfm25-recovery-4096","run_id":"lfm25-recovery-4096","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"3"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ee257ea8-aa2b-4be0-ac82-875f85fd83ef","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"intent-classification\", \"oos-calibration\", \"structured-output\"], \"configuration_id\": \"default\", \"inference_config\": {\"max_output_tokens\": 4096}, \"model_keys\": [\"lfm2.5-24b-a2b-ud-q3-k-xl\"], \"profile\": \"core\", \"run_group\": \"lfm25-recovery-4096\", \"run_id\": \"lfm25-recovery-4096\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"3\"}","model_key":null,"run_id":"lfm25-recovery-4096","sample_id":null,"status":null,"timestamp_utc":"2026-10-22T12:01:07.926009+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"481a73e1-fd9d-443c-87b5-2e8967ae354a","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"lfm25-recovery-4096\", \"run_id\": \"lfm25-recovery-4096\", \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-recovery-4096","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T12:01:07.948759+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":18285.80429102294,"error_message":null,"error_type":null,"event_id":"6161be7b-9c02-4427-bd5f-c37c66686aca","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:641822ff7e4fc678a816a765bd43258a89b949aad2eec0a36c6fb22e990053ee\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"lfm25-recovery-4096\", \"run_id\": \"lfm25-recovery-4096\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-recovery-4096","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T12:01:26.235166+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b6600a61-db0c-4bfe-8199-1592847e4f55","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:641822ff7e4fc678a816a765bd43258a89b949aad2eec0a36c6fb22e990053ee\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"lfm25-recovery-4096\", \"run_id\": \"lfm25-recovery-4096\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-recovery-4096","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T12:01:26.235608+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"5b3af2b3-1a0c-461a-ba1a-ded382f5c4d3","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3e83e05786bfa67a0e16cb28b57ec52fff1967e5e59a7bb032185b9df349aa2d\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:641822ff7e4fc678a816a765bd43258a89b949aad2eec0a36c6fb22e990053ee\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"planned_cases\": 28, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"lfm25-recovery-4096\", \"run_id\": \"lfm25-recovery-4096\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-recovery-4096","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T12:01:26.240050+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"df441ad9-0db6-4b3b-a986-1d8bcbce4a94","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:3e83e05786bfa67a0e16cb28b57ec52fff1967e5e59a7bb032185b9df349aa2d\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:641822ff7e4fc678a816a765bd43258a89b949aad2eec0a36c6fb22e990053ee\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"lfm25-recovery-4096\", \"run_id\": \"lfm25-recovery-4096\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-recovery-4096","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T12:27:38.202792+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"eb9120da-9fd5-485a-b8c8-f0fed3c88e50","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:641822ff7e4fc678a816a765bd43258a89b949aad2eec0a36c6fb22e990053ee\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"lfm25-recovery-4096\", \"run_id\": \"lfm25-recovery-4096\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-recovery-4096","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T12:27:38.205365+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d2c0d228-085c-46d5-a1f2-c53388a328a8","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:b7c8f155e579d1d1d981fb2d54369194d1a8415fe5833ed759f937156c77e621\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:641822ff7e4fc678a816a765bd43258a89b949aad2eec0a36c6fb22e990053ee\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"planned_cases\": 20, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"lfm25-recovery-4096\", \"run_id\": \"lfm25-recovery-4096\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-recovery-4096","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T12:27:38.267708+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"59b7262a-ed22-41a4-a1ac-9727c8cdc6d5","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:b7c8f155e579d1d1d981fb2d54369194d1a8415fe5833ed759f937156c77e621\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:641822ff7e4fc678a816a765bd43258a89b949aad2eec0a36c6fb22e990053ee\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"lfm25-recovery-4096\", \"run_id\": \"lfm25-recovery-4096\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-recovery-4096","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T12:48:41.084232+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6f048a73-d685-4f4d-89a7-8b22d49f42d0","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:641822ff7e4fc678a816a765bd43258a89b949aad2eec0a36c6fb22e990053ee\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"lfm25-recovery-4096\", \"run_id\": \"lfm25-recovery-4096\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-recovery-4096","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T12:48:41.084763+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"902b9ff4-df4a-47c4-a8d7-5bd891601d1a","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:c26dc80a4a3d7172c20aaf4a0d946e31de5730fcbc2b0e3ef1656e2f69c25727\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:641822ff7e4fc678a816a765bd43258a89b949aad2eec0a36c6fb22e990053ee\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"planned_cases\": 1, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"lfm25-recovery-4096\", \"run_id\": \"lfm25-recovery-4096\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-recovery-4096","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T12:48:41.159631+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"dc2d8018-8990-45c5-adff-a8c48d9500ad","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:c26dc80a4a3d7172c20aaf4a0d946e31de5730fcbc2b0e3ef1656e2f69c25727\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:641822ff7e4fc678a816a765bd43258a89b949aad2eec0a36c6fb22e990053ee\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"lfm25-recovery-4096\", \"run_id\": \"lfm25-recovery-4096\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-recovery-4096","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T12:49:48.376923+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"354fe24e-828c-4a1f-b9f8-acc280f92fae","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:641822ff7e4fc678a816a765bd43258a89b949aad2eec0a36c6fb22e990053ee\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"lfm25-recovery-4096\", \"run_id\": \"lfm25-recovery-4096\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-recovery-4096","sample_id":null,"status":"started","timestamp_utc":"2026-10-22T12:49:48.377127+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":401.68604196514934,"error_message":null,"error_type":null,"event_id":"6e0dfa44-aa73-4e34-85ec-683cf031c0e7","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:641822ff7e4fc678a816a765bd43258a89b949aad2eec0a36c6fb22e990053ee\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"provider_key\": \"korgis\", \"run_group\": \"lfm25-recovery-4096\", \"run_id\": \"lfm25-recovery-4096\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 10947100928, \"backend\": \"llama_server\", \"quantization\": \"UD-Q3_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": -1, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-recovery-4096","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T12:49:48.779012+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e42306d9-3ebb-4e57-a56d-e9e5983875af","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:3e83e05786bfa67a0e16cb28b57ec52fff1967e5e59a7bb032185b9df349aa2d\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:641822ff7e4fc678a816a765bd43258a89b949aad2eec0a36c6fb22e990053ee\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"run_group\": \"lfm25-recovery-4096\", \"run_id\": \"lfm25-recovery-4096\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-recovery-4096","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T12:49:48.790763+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a976bba2-af88-415c-9587-9770b65785df","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:b7c8f155e579d1d1d981fb2d54369194d1a8415fe5833ed759f937156c77e621\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:641822ff7e4fc678a816a765bd43258a89b949aad2eec0a36c6fb22e990053ee\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"run_group\": \"lfm25-recovery-4096\", \"run_id\": \"lfm25-recovery-4096\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-recovery-4096","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T12:49:48.792076+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6bfae070-9e11-455c-84aa-aa663050ebf4","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:c26dc80a4a3d7172c20aaf4a0d946e31de5730fcbc2b0e3ef1656e2f69c25727\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:641822ff7e4fc678a816a765bd43258a89b949aad2eec0a36c6fb22e990053ee\", \"model_key\": \"lfm2.5-24b-a2b-ud-q3-k-xl\", \"model_signature\": \"sha256:model:b548a8032a00d997901afd1e77684652fe27f55c709efdef07ef8845fd1f00ce\", \"run_group\": \"lfm25-recovery-4096\", \"run_id\": \"lfm25-recovery-4096\"}","model_key":"lfm2.5-24b-a2b-ud-q3-k-xl","run_id":"lfm25-recovery-4096","sample_id":null,"status":"completed","timestamp_utc":"2026-10-22T12:49:48.793410+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c47dbdeb-1d55-47bb-9c51-666126dfe0c2","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 20, \"completed_cases\": 49, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 49, \"run_group\": \"lfm25-recovery-4096\", \"run_id\": \"lfm25-recovery-4096\", \"skipped_cases\": 0}","model_key":null,"run_id":"lfm25-recovery-4096","sample_id":null,"status":null,"timestamp_utc":"2026-10-22T12:49:48.793583+00:00"}]},"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core":{"cells":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T11:24:59.796058+00:00","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","failure_count":0,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.577922077922078,"profile":"core","run_id":"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":3775709216,"deployment":"local","effective_model_id":"qwen3.5-2b-bf16","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","family":"qwen3.5","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","parameters_b":2.0,"provider_key":"korgis","quantization":"BF16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","model_key":"qwen3.5-2b-bf16","process_cpu_percent_avg":5.488711560059013,"process_rss_bytes_avg":5427256115.200001,"process_rss_bytes_peak":7206486016.0,"sample_count":618,"sample_interval_ms":285.9658953851292,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5988106240.0}],"run":{"completed_at_utc":"2026-10-13T11:24:59.796058+00:00","created_at_utc":"2026-10-13T11:22:34.287716+00:00","git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","integrity_state":"VALID","manifest_checksum":"1b42f62a6b25700c91835aef877bbac044ac7cb424f1687bfbb14dcd37670fdc","profile":"core","run_group":"qwen-variants-core-qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e3ac5985-c9d6-4ce1-9c14-853db4227508","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"intent-classification\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"qwen3.5-2b-bf16\"], \"profile\": \"core\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:22:34.287716+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c2729993-f6c5-40ce-9bb3-f98feb8f3393","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:22:34.306117+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":1059.4850829802454,"error_message":null,"error_type":null,"event_id":"369e7978-0420-4e0a-abad-362ac9e9aeee","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:22:35.365989+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"333b70e6-c89a-4cb5-9c22-854fe511c600","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:22:35.366324+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"af26f428-cad6-4d14-af6b-d472ebc6943b","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:22:35.372670+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"56d1381f-7798-4004-972e-e856234758e8","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:24:59.288133+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3677fb4e-1c9f-4c36-a9bc-4af83264a740","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:24:59.288390+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":494.56229200586677,"error_message":null,"error_type":null,"event_id":"bce20bc8-f857-4d59-b96c-740b0716a4ee","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:24:59.783179+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"6ccad2ff-c368-4fa2-ab53-1b296bab8983","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:24:59.795870+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b1a30ca9-d146-4958-9018-8856a353249a","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:24:59.796058+00:00"}]},"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core":{"cells":[{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T11:22:33.733080+00:00","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","failure_count":0,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.375,"profile":"core","run_id":"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":3775709216,"deployment":"local","effective_model_id":"qwen3.5-2b-bf16","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","family":"qwen3.5","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","parameters_b":2.0,"provider_key":"korgis","quantization":"BF16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","model_key":"qwen3.5-2b-bf16","process_cpu_percent_avg":6.761276471009742,"process_rss_bytes_avg":5075235362.133334,"process_rss_bytes_peak":5976489984.0,"sample_count":120,"sample_interval_ms":240.70379737386247,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":10141270016.0}],"run":{"completed_at_utc":"2026-10-13T11:22:33.733080+00:00","created_at_utc":"2026-10-13T11:22:10.289003+00:00","git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","integrity_state":"VALID","manifest_checksum":"415c471048e13237dc9465386f3381dd4c1c2e9c4abf21523692eeb0b28fb317","profile":"core","run_group":"qwen-variants-core-qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"29fe8216-86d8-4ca6-aaf4-90ee5d1b3fc5","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"qwen3.5-2b-bf16\"], \"profile\": \"core\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:22:10.289003+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"81a525ec-3337-4163-94fe-4f74969f9ea6","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:22:10.305935+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":1066.9663750450127,"error_message":null,"error_type":null,"event_id":"0abc7468-58f4-41e8-8099-1b3eb472ce2f","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:22:11.373520+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a93e0c6a-c30e-45c9-8813-3797d9ff3e65","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:22:11.373790+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3690c799-9041-4c91-a11a-5b6e4d63f1b9","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:22:11.383690+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1dd9dba0-d2b8-4e69-9bea-62c2f90ba278","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:22:33.550417+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0a3cea4a-865b-4a05-927a-e3dd8ab22ac0","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:22:33.550717+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":178.47933300072327,"error_message":null,"error_type":null,"event_id":"878fd7b8-0b3b-497d-b7a5-3afe5441d0b1","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:22:33.729443+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"45b47cf1-10e0-4ff2-95a4-7f743f54c2ff","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:22:33.732889+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"bbf61b0a-5ad6-4a36-80a3-57061c95068a","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 40, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 40, \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:22:33.733080+00:00"}]},"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core":{"cells":[{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T11:28:16.241563+00:00","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","failure_count":0,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.5,"profile":"core","run_id":"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":3775709216,"deployment":"local","effective_model_id":"qwen3.5-2b-bf16","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","family":"qwen3.5","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","parameters_b":2.0,"provider_key":"korgis","quantization":"BF16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","model_key":"qwen3.5-2b-bf16","process_cpu_percent_avg":5.117048169336818,"process_rss_bytes_avg":5994336968.810392,"process_rss_bytes_peak":9291366400.0,"sample_count":770,"sample_interval_ms":293.5350881957497,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5950177280.0}],"run":{"completed_at_utc":"2026-10-13T11:28:16.241563+00:00","created_at_utc":"2026-10-13T11:25:00.484308+00:00","git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","integrity_state":"VALID","manifest_checksum":"d430b411095f16601bc05e4f8e1aec2b0baab17a8120a3245b6bcad90fba6af1","profile":"core","run_group":"qwen-variants-core-qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"397675af-3b8c-488c-a47c-cffc7ae34439","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"oos-calibration\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"qwen3.5-2b-bf16\"], \"profile\": \"core\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:25:00.484308+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8efee649-6bbc-4d94-9847-de1eab6d1a47","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:25:00.504365+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2112.9667090135626,"error_message":null,"error_type":null,"event_id":"fa9623da-fe94-4658-92b0-315ed4985bb5","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:25:02.617759+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"19164453-eb0f-4ea8-8619-a0633d121b01","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:25:02.618066+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8249ba4d-56e4-4ff3-840a-e08517bb75f8","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:25:02.640263+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"20e31bed-ab21-411f-8382-9d89d0ecbf03","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:28:15.826537+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2efa55af-07da-47f3-9afa-c6ae612cdb24","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:28:15.826728+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":402.7898330241442,"error_message":null,"error_type":null,"event_id":"e073c203-681a-44f5-b506-e03cdd81bcd2","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:28:16.229695+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"df41f9ed-c390-4ad8-87f1-47d9533ef1ba","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:28:16.241391+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"50f1bacd-f4ec-480a-b166-cdecdf9bfec8","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:28:16.241563+00:00"}]},"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core":{"cells":[{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T11:22:09.705119+00:00","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","failure_count":0,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.6333333333333333,"profile":"core","run_id":"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":3775709216,"deployment":"local","effective_model_id":"qwen3.5-2b-bf16","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","family":"qwen3.5","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","parameters_b":2.0,"provider_key":"korgis","quantization":"BF16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","model_key":"qwen3.5-2b-bf16","process_cpu_percent_avg":5.5345916105632815,"process_rss_bytes_avg":5136314017.564444,"process_rss_bytes_peak":5894602752.0,"sample_count":226,"sample_interval_ms":279.53779863981055,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":7390167040.0}],"run":{"completed_at_utc":"2026-10-13T11:22:09.705119+00:00","created_at_utc":"2026-10-13T11:21:17.273500+00:00","git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","integrity_state":"VALID","manifest_checksum":"c991bac94a703a9917750d016bdeb1f3884e80adfd9b0234a3816622c6030601","profile":"core","run_group":"qwen-variants-core-qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"082f405b-2799-4744-bd77-70a83990eb35","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"qa-abstention\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"qwen3.5-2b-bf16\"], \"profile\": \"core\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:21:17.273500+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c6c38a0a-05fe-4139-b6a9-a7fae14e2a02","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:21:17.291619+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2133.7022919906303,"error_message":null,"error_type":null,"event_id":"e172af26-339e-46d0-a79e-669c9744f781","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:21:19.425849+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f7c81657-f04f-4e2f-94b9-6000d6859e79","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:21:19.426144+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a8bfb5a3-99ee-496a-a626-d8ac8a4e5717","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:21:19.450886+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"bf1e03c2-d7de-45fa-a37a-b159fef490ea","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:22:09.514218+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b299fc55-efdb-4980-a362-b64bfa9a3967","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:22:09.514426+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":184.888125048019,"error_message":null,"error_type":null,"event_id":"66c04c3f-46b1-489b-b3e6-450e09cd77d4","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:22:09.699493+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3a434605-4591-4707-abef-9eda77f07d0d","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:22:09.704905+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"716ecc22-fd88-406a-9b04-c1d98c0e6ab6","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:22:09.705119+00:00"}]},"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core":{"cells":[{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T11:21:16.650276+00:00","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","failure_count":0,"model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7166666666666667,"profile":"core","run_id":"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":3775709216,"deployment":"local","effective_model_id":"qwen3.5-2b-bf16","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","family":"qwen3.5","model_id":"unsloth/qwen3.5-2b-bf16","model_key":"qwen3.5-2b-bf16","model_signature":"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495","parameters_b":2.0,"provider_key":"korgis","quantization":"BF16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76","model_key":"qwen3.5-2b-bf16","process_cpu_percent_avg":5.386938078405328,"process_rss_bytes_avg":4179186975.941509,"process_rss_bytes_peak":4437606400.0,"sample_count":391,"sample_interval_ms":306.911332594406,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":4923883520.0}],"run":{"completed_at_utc":"2026-10-13T11:21:16.650276+00:00","created_at_utc":"2026-10-13T11:19:09.277950+00:00","git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","integrity_state":"VALID","manifest_checksum":"0ec539a96736fd76813dbf27048a89c1d949a75a13f827c93e344f3784af194f","profile":"core","run_group":"qwen-variants-core-qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"4900f224-1548-4b30-bdd4-15b728caf2ea","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"structured-output\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"qwen3.5-2b-bf16\"], \"profile\": \"core\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:19:09.277950+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"028d1c4f-c9d3-4624-b4a7-693f7b344424","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:19:09.295117+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":20316.69658399187,"error_message":null,"error_type":null,"event_id":"b30d5fb1-b9eb-4065-a134-d447f760fb19","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:19:29.612582+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f3b6fbfb-9e2c-4bd9-bdf7-ee7651e58ffc","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:19:29.614127+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"9818638b-da04-411d-be27-d17aeb245838","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:19:29.664457+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"b6c9cff6-d039-40ff-9021-a3da0339880f","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:21:16.408490+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"78623538-f18c-4582-aa57-ea9233f0e726","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:21:16.408712+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":236.26958299428225,"error_message":null,"error_type":null,"event_id":"3d55bf16-0e7c-4eb0-b03b-cdf6f950cf40","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 3775709216, \"backend\": \"llama_server\", \"quantization\": \"BF16\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-2b-bf16\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:21:16.645139+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"79d6c304-fc67-49c2-9f6a-40cdbd113444","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:d2c23feabdff30dcb8525be7b861f20fe5cae29acdeb79ab5f755554a1934c76\", \"model_key\": \"qwen3.5-2b-bf16\", \"model_signature\": \"sha256:model:111f71d938f753cbbfe5cfc22e005b9a26c1c7ad6b1bde3bd5c7a039f37db495\", \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core\"}","model_key":"qwen3.5-2b-bf16","run_id":"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:21:16.650134+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"5afe8c0a-81ab-444f-a7cd-f989d3394112","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"qwen-variants-core-qwen3.5-2b-bf16\", \"run_id\": \"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-2b-bf16-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:21:16.650276+00:00"}]},"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core":{"cells":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T11:37:13.963536+00:00","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","failure_count":0,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6493506493506493,"profile":"core","run_id":"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":4482403488,"deployment":"local","effective_model_id":"qwen3.5-4b-q8","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","family":"qwen3.5","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","parameters_b":4.0,"provider_key":"korgis","quantization":"Q8_0","run_id":"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","model_key":"qwen3.5-4b-q8","process_cpu_percent_avg":5.113588871923991,"process_rss_bytes_avg":7039335235.285094,"process_rss_bytes_peak":10390700032.0,"sample_count":1041,"sample_interval_ms":311.43875954839615,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5581979648.0}],"run":{"completed_at_utc":"2026-10-13T11:37:13.963536+00:00","created_at_utc":"2026-10-13T11:32:23.476561+00:00","git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","integrity_state":"VALID","manifest_checksum":"e6dd7e82e133b30ea869a90ebf53be0e9769814391279970fd85763de3ff2dc0","profile":"core","run_group":"qwen-variants-core-qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"7a42d5ab-ed15-4e1a-804b-1e818942d23a","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"intent-classification\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"qwen3.5-4b-q8\"], \"profile\": \"core\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:32:23.476561+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"461b7d18-e45b-4817-9bc8-5d0a27c57408","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:32:23.494365+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2083.895916992333,"error_message":null,"error_type":null,"event_id":"538c25c6-13c6-4fb1-bc6d-73202f8ca611","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:32:25.578730+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"22dfdaf0-944c-4cb5-9f41-c5f596689ea6","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:32:25.579152+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"24bcd25a-79b5-4080-ac2a-2e70e1095583","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:32:25.586658+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a86aa91a-505d-4dcf-96e7-828441b875ac","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:37:13.550206+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"33f47832-2020-485a-b759-c4b15bcf5961","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:37:13.550404+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":393.20395898539573,"error_message":null,"error_type":null,"event_id":"d73886cf-6eca-4737-a346-7746f6a35894","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:37:13.943756+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"bc1278f1-caed-49f8-b3a9-71c56ab69c1a","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:37:13.963262+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"4edaf0fd-47c9-433a-b63a-4f764259f51a","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-4b-q8-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:37:13.963536+00:00"}]},"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core":{"cells":[{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T11:32:22.918386+00:00","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","failure_count":0,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.5,"profile":"core","run_id":"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":4482403488,"deployment":"local","effective_model_id":"qwen3.5-4b-q8","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","family":"qwen3.5","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","parameters_b":4.0,"provider_key":"korgis","quantization":"Q8_0","run_id":"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","model_key":"qwen3.5-4b-q8","process_cpu_percent_avg":6.982037738154942,"process_rss_bytes_avg":7516327697.066668,"process_rss_bytes_peak":9962160128.0,"sample_count":132,"sample_interval_ms":291.2576532937237,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":11326062592.0}],"run":{"completed_at_utc":"2026-10-13T11:32:22.918386+00:00","created_at_utc":"2026-10-13T11:31:51.260611+00:00","git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","integrity_state":"VALID","manifest_checksum":"ad42bd827e985d326c9750efa563a597c8d331a84e46474d8cfd9f49427838f0","profile":"core","run_group":"qwen-variants-core-qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"78ffb77d-c4b9-4d39-9b93-f22a2b817d2d","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"qwen3.5-4b-q8\"], \"profile\": \"core\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:31:51.260611+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"07e605f7-8b7f-4c5d-b764-219abef66b45","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:31:51.278338+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2097.4110839888453,"error_message":null,"error_type":null,"event_id":"75f6e0ad-df1a-4f90-8305-1b1c61a75e68","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:31:53.376184+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"11c34582-e93d-40f9-b58b-12b97a22efec","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:31:53.376485+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"eda8a694-b8e8-4138-a104-e66fe37e14e2","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:31:53.388247+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f090b64a-be50-47c1-83dc-d46e6adb7369","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:32:22.729789+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a1ace6bb-4b20-4321-9bb5-45427c650d7e","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:32:22.730108+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":184.28400001721457,"error_message":null,"error_type":null,"event_id":"7da54d5f-25ee-4205-9860-ec20f6aafc64","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:32:22.914672+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d4d152cc-8939-49d7-b5c7-2abc0082eae9","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:32:22.918201+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"603d115b-3613-452c-b410-246feb2fac3e","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 40, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 40, \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-4b-q8-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:32:22.918386+00:00"}]},"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core":{"cells":[{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T11:43:07.482305+00:00","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","failure_count":0,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.9805194805194806,"profile":"core","run_id":"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":4482403488,"deployment":"local","effective_model_id":"qwen3.5-4b-q8","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","family":"qwen3.5","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","parameters_b":4.0,"provider_key":"korgis","quantization":"Q8_0","run_id":"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","model_key":"qwen3.5-4b-q8","process_cpu_percent_avg":5.610485336009209,"process_rss_bytes_avg":7096272805.442177,"process_rss_bytes_peak":11637391360.0,"sample_count":1211,"sample_interval_ms":319.76690310093494,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5589811200.0}],"run":{"completed_at_utc":"2026-10-13T11:43:07.482305+00:00","created_at_utc":"2026-10-13T11:37:14.796112+00:00","git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","integrity_state":"VALID","manifest_checksum":"624640de84cb3cae532e9c40922cb02bd72de7f952745bf7518c00bdbbd8aa35","profile":"core","run_group":"qwen-variants-core-qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"134a88a5-d597-4d07-a3f1-c5e33fff43bc","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"oos-calibration\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"qwen3.5-4b-q8\"], \"profile\": \"core\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:37:14.796112+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"e87e689e-41ae-4da3-a4bd-84205d714ae8","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:37:14.818079+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2088.7134579825215,"error_message":null,"error_type":null,"event_id":"9ca8e252-7ce3-477a-bed4-744015815a42","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:37:16.907240+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"72220554-90aa-4727-9be6-b4c9ec07764b","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:37:16.907521+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"57f7c035-bc88-4df0-81e4-02aae98f11eb","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:37:16.929363+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0a23da95-cfa2-40a9-bc0f-8985fe749092","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:43:07.018845+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ac714686-037a-466b-84eb-6b869321279b","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:43:07.019113+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":448.33962502889335,"error_message":null,"error_type":null,"event_id":"d4f59af9-eaf7-4c4b-8e71-8aeb51e62443","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:43:07.467689+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"49604468-95c8-4f96-874d-ae466e65a674","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:43:07.482109+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"ced442f3-52dd-4839-ace5-be8b8d77dcb4","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-4b-q8-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:43:07.482305+00:00"}]},"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core":{"cells":[{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T11:31:50.661853+00:00","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","failure_count":0,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"profile":"core","run_id":"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":4482403488,"deployment":"local","effective_model_id":"qwen3.5-4b-q8","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","family":"qwen3.5","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","parameters_b":4.0,"provider_key":"korgis","quantization":"Q8_0","run_id":"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","model_key":"qwen3.5-4b-q8","process_cpu_percent_avg":6.023120834777095,"process_rss_bytes_avg":7696510827.330371,"process_rss_bytes_peak":10978181120.0,"sample_count":278,"sample_interval_ms":297.275809231578,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":7347224576.0}],"run":{"completed_at_utc":"2026-10-13T11:31:50.661853+00:00","created_at_utc":"2026-10-13T11:30:39.250038+00:00","git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","integrity_state":"VALID","manifest_checksum":"4ab7bc2b0704a47bfea448accd86a84b1eca5c2a578838db7f82a80b2e3aaba4","profile":"core","run_group":"qwen-variants-core-qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8d3a2391-4d92-4f88-8122-9d8e684b3adb","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"qa-abstention\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"qwen3.5-4b-q8\"], \"profile\": \"core\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:30:39.250038+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"90c583ed-633c-4a21-9120-5b058c4e7e7e","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:30:39.266830+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2099.1528329905123,"error_message":null,"error_type":null,"event_id":"7ef1059c-5947-481f-b420-1e017a011d85","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:30:41.366376+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"11e8c978-bd72-4635-a1ec-9066837b229c","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:30:41.366706+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8316ec57-db93-49ad-9297-425263e996e2","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:30:41.393870+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3b16f9c7-6958-47a8-9d07-86f25eb121b5","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:31:50.363072+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"cbade5a8-35fe-4cca-be89-67984f42868f","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:31:50.363236+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":291.8584590079263,"error_message":null,"error_type":null,"event_id":"502ba657-b7f6-4cc5-8854-699f0fda48f2","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:31:50.655243+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"25616996-215d-4460-ab40-5bc6407ed973","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:31:50.661711+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"07f4254f-b09b-4149-9618-24074a5a9681","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-4b-q8-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:31:50.661853+00:00"}]},"qwen-variants-core-qwen3.5-4b-q8-structured-output-core":{"cells":[{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T11:30:38.611895+00:00","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","failure_count":0,"model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7333333333333333,"profile":"core","run_id":"qwen-variants-core-qwen3.5-4b-q8-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"qwen-variants-core-qwen3.5-4b-q8-structured-output-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":4482403488,"deployment":"local","effective_model_id":"qwen3.5-4b-q8","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","family":"qwen3.5","model_id":"unsloth/qwen3.5-4b-q8","model_key":"qwen3.5-4b-q8","model_signature":"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3","parameters_b":4.0,"provider_key":"korgis","quantization":"Q8_0","run_id":"qwen-variants-core-qwen3.5-4b-q8-structured-output-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895","model_key":"qwen3.5-4b-q8","process_cpu_percent_avg":5.698576220486698,"process_rss_bytes_avg":6661104577.102827,"process_rss_bytes_peak":9700769792.0,"sample_count":477,"sample_interval_ms":317.01940462127334,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5716574208.0}],"run":{"completed_at_utc":"2026-10-13T11:30:38.611895+00:00","created_at_utc":"2026-10-13T11:28:17.184933+00:00","git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","integrity_state":"VALID","manifest_checksum":"75f0531efcdf87461adb6cbf02e79a9e9076b9c7b36a10612faaa0b076183b7a","profile":"core","run_group":"qwen-variants-core-qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-structured-output-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"022d7506-a61f-4389-8013-9e518abf1084","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"structured-output\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"qwen3.5-4b-q8\"], \"profile\": \"core\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-structured-output-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-4b-q8-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:28:17.184933+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"3c825223-8ea6-4a64-8c28-c712a9563f01","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:28:17.205320+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":4114.236207969952,"error_message":null,"error_type":null,"event_id":"d8fef1cd-9722-4c58-9bbd-506b9acef882","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:28:21.320084+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"07dc7ba9-1872-4e32-971f-8754588f742a","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:28:21.320405+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f2aae76a-f91d-43e8-bebd-d8bafbd41d34","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:28:21.371241+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"bec6ba68-14b4-4516-884d-86da434250b5","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:30:38.210762+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"a0e37896-4e7e-4110-bf4f-10b647761ae9","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:30:38.210967+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":394.31983296526596,"error_message":null,"error_type":null,"event_id":"57f4fcdd-72d6-46a5-a223-f2d52c7bc300","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4482403488, \"backend\": \"llama_server\", \"quantization\": \"Q8_0\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-4b-q8\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:30:38.605455+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f0d303ab-dbd2-435a-b9ba-b54b2812d585","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:c197178d81afed14ff4056bbf995e44c2d9884cae2336ea223c49cf99bb44895\", \"model_key\": \"qwen3.5-4b-q8\", \"model_signature\": \"sha256:model:e07111adfd3068113b4bcc63a9ba61cfd34f34ea6d9469777ef977d8f6df2bb3\", \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-structured-output-core\"}","model_key":"qwen3.5-4b-q8","run_id":"qwen-variants-core-qwen3.5-4b-q8-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:30:38.611753+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"32c72bb5-fff8-4b90-9676-c43354a9aa9a","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"qwen-variants-core-qwen3.5-4b-q8\", \"run_id\": \"qwen-variants-core-qwen3.5-4b-q8-structured-output-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-4b-q8-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:30:38.611895+00:00"}]},"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core":{"cells":[{"benchmark_signature":"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b","capability_id":"intent-classification","comparison_metric":"accuracy","completed_at_utc":"2026-10-13T11:55:33.171703+00:00","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","failure_count":0,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","practical_delta":0.03,"primary_metric":"accuracy","primary_value":0.6883116883116883,"profile":"core","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","sample_count":154,"status":"COMPLETED","task_id":"intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":4121781472,"deployment":"local","effective_model_id":"qwen3.5-9b-ud-q2kxl","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","family":"qwen3.5","model_id":"qwen/qwen3.5-9b-ud-q2kxl","model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","parameters_b":9.0,"provider_key":"korgis","quantization":"UD-Q2_K_XL","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"intent-classification","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","model_key":"qwen3.5-9b-ud-q2kxl","process_cpu_percent_avg":3.288803542984178,"process_rss_bytes_avg":6954068738.250033,"process_rss_bytes_peak":11224416256.0,"sample_count":1539,"sample_interval_ms":314.5480896344338,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5547704320.0}],"run":{"completed_at_utc":"2026-10-13T11:55:33.171703+00:00","created_at_utc":"2026-10-13T11:48:03.877002+00:00","git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","integrity_state":"VALID","manifest_checksum":"61f91f98538768f5db24dde01d31e36cfbaf5637b6f8566fb12c454222b12697","profile":"core","run_group":"qwen-variants-core-qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"0772cc07-e7e7-4a23-b9b0-8950fdaf9732","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"intent-classification\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"qwen3.5-9b-ud-q2kxl\"], \"profile\": \"core\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:48:03.877002+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f90f563b-30d1-4dd9-9c2c-9722f3c145d7","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:48:03.895818+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2067.285166995134,"error_message":null,"error_type":null,"event_id":"ee29ade9-e376-45ab-9320-9446be938c45","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:48:05.963526+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c549a2a9-0da6-4e8a-9c9b-1f6b7a99998f","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:48:05.963811+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1f3e8a14-a9fa-46fc-9330-051c7a76d182","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:48:05.970222+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"5bc7b4bb-0121-4490-939c-c240fc5f70e4","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:55:32.763513+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"cec84ad8-c697-4e79-afc9-fb6cc890780c","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:55:32.763754+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":397.73949998198077,"error_message":null,"error_type":null,"event_id":"870c3afa-9322-4d45-be33-029952cc5f81","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:55:33.161677+00:00"},{"attempt":null,"capability_id":"intent-classification","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"4dbc6ec2-53ff-480a-9dc9-106f05b3da26","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:9ed670e7bf347a304b5a8e0f66561d1ef0fa44378a6ac03e3a663272121c3f0b\", \"capability_id\": \"intent-classification\", \"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:55:33.171551+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f1f365b1-b62e-4111-8db0-c1b892bb97a3","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-intent-classification-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:55:33.171703+00:00"}]},"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core":{"cells":[{"benchmark_signature":"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1","capability_id":"mathematical-reasoning-v2","comparison_metric":"final_answer_accuracy","completed_at_utc":"2026-10-13T11:48:03.359698+00:00","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","failure_count":0,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","practical_delta":0.05,"primary_metric":"final_answer_accuracy","primary_value":0.55,"profile":"core","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core","sample_count":40,"status":"COMPLETED","task_id":"mathematical-reasoning"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":4121781472,"deployment":"local","effective_model_id":"qwen3.5-9b-ud-q2kxl","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","family":"qwen3.5","model_id":"qwen/qwen3.5-9b-ud-q2kxl","model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","parameters_b":9.0,"provider_key":"korgis","quantization":"UD-Q2_K_XL","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"mathematical-reasoning-v2","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","model_key":"qwen3.5-9b-ud-q2kxl","process_cpu_percent_avg":4.913551711729864,"process_rss_bytes_avg":7164752896.0,"process_rss_bytes_peak":9602973696.0,"sample_count":164,"sample_interval_ms":288.91353003637056,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":12627738624.0}],"run":{"completed_at_utc":"2026-10-13T11:48:03.359698+00:00","created_at_utc":"2026-10-13T11:47:22.283975+00:00","git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","integrity_state":"VALID","manifest_checksum":"3ee9cd58168d5965209f3d86938fe6f595ba7e28f8594711f3eee7be81047ce9","profile":"core","run_group":"qwen-variants-core-qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"c7af8b00-2b9b-4525-90fc-208b1f3a0f3d","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"mathematical-reasoning\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"qwen3.5-9b-ud-q2kxl\"], \"profile\": \"core\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:47:22.283975+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"54cc412a-adee-494d-88fb-ef54fa8f4b71","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:47:22.303860+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2103.983542008791,"error_message":null,"error_type":null,"event_id":"bf1d6709-0b16-4c6b-8065-e3fa5a4c7d35","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:47:24.408283+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"bc6db778-d8a4-48ae-ab56-76e05c5db042","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:47:24.408633+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"5386e8bc-a65a-4763-962f-7d254507720d","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"planned_cases\": 40, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:47:24.420981+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2fd6da69-1c0a-4f12-9c82-8a858007f1e8","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:48:03.175100+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2e58477f-34f0-4ddb-b1c3-82bb8d176f7e","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:48:03.175344+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":180.4175419965759,"error_message":null,"error_type":null,"event_id":"6d0a93a8-fe2c-4655-a533-345d2a778000","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:48:03.355957+00:00"},{"attempt":null,"capability_id":"mathematical-reasoning-v2","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f1143668-00b8-45a2-b1fc-4f1d5ddcaa8a","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:2c517d2bdadd3d3525dbe397c2b0f47388992caec0ce151a3a514c119fae23b1\", \"capability_id\": \"mathematical-reasoning\", \"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:48:03.359552+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"503a883e-13a2-4146-b071-bac8d6a86baa","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 40, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 40, \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-mathematical-reasoning-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:48:03.359698+00:00"}]},"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core":{"cells":[{"benchmark_signature":"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2","capability_id":"oos-calibration","comparison_metric":"oos_correct","completed_at_utc":"2026-10-13T12:04:10.873473+00:00","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","failure_count":0,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","practical_delta":0.04,"primary_metric":"oos_detection_rate","primary_value":0.961038961038961,"profile":"core","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core","sample_count":154,"status":"COMPLETED","task_id":"calibrated-intent-classification"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":4121781472,"deployment":"local","effective_model_id":"qwen3.5-9b-ud-q2kxl","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","family":"qwen3.5","model_id":"qwen/qwen3.5-9b-ud-q2kxl","model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","parameters_b":9.0,"provider_key":"korgis","quantization":"UD-Q2_K_XL","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"oos-calibration","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","model_key":"qwen3.5-9b-ud-q2kxl","process_cpu_percent_avg":3.489486960210178,"process_rss_bytes_avg":6127369651.484418,"process_rss_bytes_peak":9636659200.0,"sample_count":1713,"sample_interval_ms":322.34991046526653,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5265096704.0}],"run":{"completed_at_utc":"2026-10-13T12:04:10.873473+00:00","created_at_utc":"2026-10-13T11:55:33.980420+00:00","git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","integrity_state":"VALID","manifest_checksum":"443ce6dff80bd2d6f63c61459e01376ab02c0bffb89d847ecfb91cfe46f0eeb1","profile":"core","run_group":"qwen-variants-core-qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8f163e90-f0db-4bea-a8b5-ff82231f0c49","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"oos-calibration\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"qwen3.5-9b-ud-q2kxl\"], \"profile\": \"core\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:55:33.980420+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f1bc0235-b9da-44af-b692-a489ce9c052d","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:55:33.999035+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2090.447458962444,"error_message":null,"error_type":null,"event_id":"78a65928-0f37-4a5d-948d-3bbd398bd286","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:55:36.089912+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"5a567c51-dabc-4d24-9d70-cc327f037c18","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:55:36.090139+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"9b4a52da-6f13-48b0-bea6-99474f24c305","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"planned_cases\": 154, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:55:36.112433+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"cbb078a1-4fa3-4d71-b0fc-d05df8601034","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T12:04:10.442204+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f7580838-f997-4817-80dc-3885d0ce2c1e","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T12:04:10.442773+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":401.6117079881951,"error_message":null,"error_type":null,"event_id":"91c5c54d-8223-4274-bcc0-7a908f8eca8c","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T12:04:10.844566+00:00"},{"attempt":null,"capability_id":"oos-calibration","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"195357a8-5370-4191-8a2a-13a00e6fd76b","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:da0de015926b16b30189f7256a762b3ee8912284de39a953fb5f877f476f12f2\", \"capability_id\": \"oos-calibration\", \"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T12:04:10.873307+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"f46e4ea4-ae1b-4b08-84ee-b98cc0db7ea7","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 154, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 154, \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-oos-calibration-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T12:04:10.873473+00:00"}]},"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core":{"cells":[{"benchmark_signature":"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399","capability_id":"qa-abstention","comparison_metric":"qa_correct","completed_at_utc":"2026-10-13T11:47:21.635679+00:00","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","failure_count":0,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","practical_delta":0.05,"primary_metric":"qa_correct","primary_value":0.8666666666666667,"profile":"core","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core","sample_count":60,"status":"COMPLETED","task_id":"qa-abstention"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":4121781472,"deployment":"local","effective_model_id":"qwen3.5-9b-ud-q2kxl","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","family":"qwen3.5","model_id":"qwen/qwen3.5-9b-ud-q2kxl","model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","parameters_b":9.0,"provider_key":"korgis","quantization":"UD-Q2_K_XL","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"qa-abstention","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","model_key":"qwen3.5-9b-ud-q2kxl","process_cpu_percent_avg":4.725490643337322,"process_rss_bytes_avg":7851091423.1669855,"process_rss_bytes_peak":11102388224.0,"sample_count":336,"sample_interval_ms":289.6220233097523,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":8726102016.0}],"run":{"completed_at_utc":"2026-10-13T11:47:21.635679+00:00","created_at_utc":"2026-10-13T11:45:54.649608+00:00","git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","integrity_state":"VALID","manifest_checksum":"57c1ba0a111fe5e56c0d1763240837aa05a20f3e20f0abe929a856794e25ffb5","profile":"core","run_group":"qwen-variants-core-qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"dea81079-13ad-4500-9a6c-049cac66104d","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"qa-abstention\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"qwen3.5-9b-ud-q2kxl\"], \"profile\": \"core\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:45:54.649608+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"5752ac7a-851a-4526-b082-45cf3ca6b4e5","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:45:54.668363+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":2095.6244170083664,"error_message":null,"error_type":null,"event_id":"6916c4c8-bc13-4bfe-bc72-6f610b37374f","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:45:56.764477+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"712be4ca-366f-495a-93fd-292e5f77e483","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:45:56.764716+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"17f53835-53d0-48eb-81d1-72eaeb4b660e","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:45:56.787096+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"d31ea196-ce6b-444b-95aa-9f6104e1b775","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:47:21.281999+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"07aaad00-e44f-4ff2-a20a-eacadd2fbaad","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:47:21.282391+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":346.18612501071766,"error_message":null,"error_type":null,"event_id":"618ebf45-34ff-41ac-a58e-b3779b811b65","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:47:21.628947+00:00"},{"attempt":null,"capability_id":"qa-abstention","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"318324c8-aaaf-4664-98ac-4f7cb73d9bb5","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 6, \"benchmark_signature\": \"sha256:benchmark:892e6c3f5346792827c489caa4aca543b605777c43aa2d45e808eb7926ca5399\", \"capability_id\": \"qa-abstention\", \"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:47:21.635525+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"1acf025d-8ff3-4e1e-859f-7982556e347b","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 6, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-qa-abstention-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:47:21.635679+00:00"}]},"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core":{"cells":[{"benchmark_signature":"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1","capability_id":"structured-output","comparison_metric":"exact_match","completed_at_utc":"2026-10-13T11:45:53.885678+00:00","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","failure_count":0,"model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","practical_delta":0.05,"primary_metric":"exact_match","primary_value":0.7166666666666667,"profile":"core","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core","sample_count":60,"status":"COMPLETED","task_id":"structured-output"}],"configuration":{"changed_dimension":null,"configuration_id":"default","experiment_kind":"standard","inference_json":"{}","is_baseline":false,"label":null,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core","runtime_json":"{}","sweep_id":null},"failure_summary":[],"models":[{"artifact_format":"gguf","artifact_size_bytes":4121781472,"deployment":"local","effective_model_id":"qwen3.5-9b-ud-q2kxl","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","family":"qwen3.5","model_id":"qwen/qwen3.5-9b-ud-q2kxl","model_key":"qwen3.5-9b-ud-q2kxl","model_signature":"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2","parameters_b":9.0,"provider_key":"korgis","quantization":"UD-Q2_K_XL","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core","runtime_key":"korgis-local"}],"resources":[{"accelerator_memory_bytes_peak":null,"capability_id":"structured-output","execution_signature":"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519","model_key":"qwen3.5-9b-ud-q2kxl","process_cpu_percent_avg":4.974081455251981,"process_rss_bytes_avg":7070526205.54296,"process_rss_bytes_peak":10778755072.0,"sample_count":559,"sample_interval_ms":311.91939930876663,"sampling_error_count":0,"scope":"owned_backend_process","source":"[redacted local or credential-bearing value]","system_available_memory_bytes_min":5600329728.0}],"run":{"completed_at_utc":"2026-10-13T11:45:53.885678+00:00","created_at_utc":"2026-10-13T11:43:08.208319+00:00","git_commit":"7f5324478e246f94f93ee8a5a32462d44851f538","integrity_state":"VALID","manifest_checksum":"c970a0285108eb8ef410a393b31228a5583309452fe379bd94fca8d793fa7540","profile":"core","run_group":"qwen-variants-core-qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core","seed":42,"source_dir":null,"status":"COMPLETED","suite_id":"capability-core","suite_version":"2"},"schema_version":"1","timeline":[{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"2c201d66-1052-4eb5-98f0-902922bf3112","event_type":"run.started","metadata_json":"{\"capability_ids\": [\"structured-output\"], \"configuration_id\": \"default\", \"inference_config\": {}, \"model_keys\": [\"qwen3.5-9b-ud-q2kxl\"], \"profile\": \"core\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core\", \"runtime_config\": {}, \"seed\": 42, \"suite_id\": \"capability-core\", \"suite_version\": \"2\"}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:43:08.208319+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"08167962-96c7-4e07-9ad4-7321a2945729","event_type":"model.prepare.started","metadata_json":"{\"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core\", \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:43:08.227185+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":4096.867709013168,"error_message":null,"error_type":null,"event_id":"9d89d27b-f4be-45bc-9911-daa46058e9e8","event_type":"model.prepare.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:43:12.324496+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"63c7efd9-dadd-40c4-b758-ea87de608901","event_type":"capability.loading.started","metadata_json":"{\"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:43:12.324788+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"dc98760f-de53-4e6d-9517-ef6206b10029","event_type":"capability.started","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"planned_cases\": 60, \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:43:12.376221+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"07d3a1df-bde4-443d-be57-0b91a0702082","event_type":"capability.completed","metadata_json":"{\"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"profile\": \"core\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:45:53.475798+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8d89a2fb-61a2-4e67-b758-b9b374bb4a79","event_type":"model.release.started","metadata_json":"{\"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core","sample_id":null,"status":"started","timestamp_utc":"2026-10-13T11:45:53.475979+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":401.5975419897586,"error_message":null,"error_type":null,"event_id":"f19eee0a-a24e-44c0-9ec2-49df1990fd5f","event_type":"model.release.completed","metadata_json":"{\"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"provider_key\": \"korgis\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core\", \"runtime_execution_metadata\": {\"artifact_size_bytes\": 4121781472, \"backend\": \"llama_server\", \"quantization\": \"UD-Q2_K_XL\", \"runtime_config\": {\"backend\": \"llama_server\", \"ctx_size\": 8192, \"enable_thinking\": false, \"flash_attn\": true, \"max_kv_size\": null, \"n_batch\": 512, \"n_gpu_layers\": 0, \"n_threads\": 8, \"n_ubatch\": 512, \"offload_kqv\": true, \"show_thinking\": false, \"timeout\": 1200, \"use_mmap\": true}, \"runtime_key\": \"qwen3.5-9b-ud-q2kxl\", \"runtime_source\": \"korgis\"}, \"runtime_key\": \"korgis-local\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:45:53.877739+00:00"},{"attempt":null,"capability_id":"structured-output","case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"cf70e6e7-a14e-4536-8e26-e87f21444eb2","event_type":"capability.aggregated","metadata_json":"{\"aggregate_count\": 7, \"benchmark_signature\": \"sha256:benchmark:8e4a417e91c378014ebec9604bce4c41ff631f4521da1d5a52ca5e451118a4a1\", \"capability_id\": \"structured-output\", \"execution_signature\": \"sha256:execution:cd1f305d12290c8876ae51ede9cdcf77e484bf49aacbae7597405bab9f818519\", \"model_key\": \"qwen3.5-9b-ud-q2kxl\", \"model_signature\": \"sha256:model:699302e2aa5f01b4eb6102a1011e1a74dc17b41392b003f53ae05cb18e7388c2\", \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core\"}","model_key":"qwen3.5-9b-ud-q2kxl","run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core","sample_id":null,"status":"completed","timestamp_utc":"2026-10-13T11:45:53.885502+00:00"},{"attempt":null,"capability_id":null,"case_id":null,"dataset_id":null,"duration_ms":null,"error_message":null,"error_type":null,"event_id":"8f0d184e-59ad-49a1-a724-1f33086f8eef","event_type":"run.completed","metadata_json":"{\"aggregate_count\": 7, \"completed_cases\": 60, \"failed_cases\": 0, \"model_failures\": 0, \"planned_cases\": 60, \"run_group\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl\", \"run_id\": \"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core\", \"skipped_cases\": 0}","model_key":null,"run_id":"qwen-variants-core-qwen3.5-9b-ud-q2kxl-structured-output-core","sample_id":null,"status":null,"timestamp_utc":"2026-10-13T11:45:53.885678+00:00"}]}},"schema_version":"1"} diff --git a/experiments/model-capability-benchmark/src/model_capability_bench/analytics/dashboard_export.py b/experiments/model-capability-benchmark/src/model_capability_bench/analytics/dashboard_export.py index 79fc0c6c..d6ba6b43 100644 --- a/experiments/model-capability-benchmark/src/model_capability_bench/analytics/dashboard_export.py +++ b/experiments/model-capability-benchmark/src/model_capability_bench/analytics/dashboard_export.py @@ -9,6 +9,7 @@ build_decision_overview, ) from model_capability_bench.analytics.pareto import build_frontier_payload +from model_capability_bench.analytics.public_bundle import write_public_bundle from model_capability_bench.analytics.sensitivity import build_sensitivity_payload from model_capability_bench.runner.comparison import paired_binary_comparison @@ -40,7 +41,12 @@ def export_dashboard_data( *, database_path: Path, output_dir: Path, + result_view: str = "latest", + public_output_path: Path | None = None, ) -> dict[str, str]: + if result_view not in {"latest", "standard"}: + raise ValueError("result_view must be latest or standard") + quality_view = "v_latest_quality_results" if result_view == "latest" else "v_current_quality_results" database_path = database_path.resolve() output_dir = output_dir.resolve() output_dir.mkdir(parents=True, exist_ok=True) @@ -49,7 +55,7 @@ def export_dashboard_data( try: current = _rows( connection, - """ + f""" SELECT q.run_id, q.model_key, @@ -76,13 +82,16 @@ def export_dashboard_data( m.artifact_format, m.family, m.parameters_b, - m.artifact_size_bytes - FROM v_current_quality_results q + m.artifact_size_bytes, + rc.experiment_kind, + rc.inference_json AS generation_policy + FROM {quality_view} q JOIN models m ON m.run_id = q.run_id AND m.model_key = q.model_key JOIN runs r ON r.run_id = q.run_id + LEFT JOIN run_configurations rc ON rc.run_id = q.run_id ORDER BY q.capability_id, q.model_key """, ) @@ -111,6 +120,7 @@ def export_dashboard_data( for row in current_resources } for row in current: + row["generation_policy"] = _decode_json(row.get("generation_policy")) row["resource_summary"] = resources_by_cell.get( ( str(row["run_id"]), @@ -168,6 +178,12 @@ def export_dashboard_data( sensitivity = build_sensitivity_payload(connection) overview = { "schema_version": "2", + "result_view": result_view, + "consolidations": [ + {"run_id": row["run_id"], "model_key": row["model_key"], + "capability_id": row["capability_id"], "generation_policy": row["generation_policy"]} + for row in current if row.get("experiment_kind") == "consolidated" + ], "models": list(models.values()), "capabilities": capabilities, "cells": current, @@ -214,7 +230,7 @@ def export_dashboard_data( sig = benchmark_signatures[0] case_rows = _rows( connection, - """ + f""" SELECT c.run_id, c.case_id, @@ -245,7 +261,7 @@ def export_dashboard_data( AND c.benchmark_signature = ? AND c.run_id IN ( SELECT run_id - FROM v_current_quality_results + FROM {quality_view} WHERE capability_id = ? AND benchmark_signature = ? ) @@ -589,7 +605,15 @@ def comparison_evidence( finally: connection.close() + if public_output_path is None: + public_output_path = ( + output_dir.parent.parent / "published" / "dashboard-data.json" + if output_dir.name == "dashboard" and output_dir.parent.name == "analytics" + else output_dir / "dashboard-data.json" + ) + bundle_path = write_public_bundle(output_dir, destination=public_output_path) return { "index": str(output_dir / "index.json"), "overview": str(output_dir / "overview.json"), + "bundle": str(bundle_path), } diff --git a/experiments/model-capability-benchmark/src/model_capability_bench/analytics/decision_summary.py b/experiments/model-capability-benchmark/src/model_capability_bench/analytics/decision_summary.py index 7737e157..d089fb74 100644 --- a/experiments/model-capability-benchmark/src/model_capability_bench/analytics/decision_summary.py +++ b/experiments/model-capability-benchmark/src/model_capability_bench/analytics/decision_summary.py @@ -704,8 +704,15 @@ def build_decision_overview( "artifact_format": cell.get("artifact_format") or artifact.get("format"), "artifact_size_bytes": artifact.get("size_bytes"), "tags": cfg.get("tags") or [], + "result_policy": ( + "consolidated-latest-attempt" + if cell.get("experiment_kind") == "consolidated" else "standard" + ), } + if cell.get("experiment_kind") == "consolidated": + model_meta[model_key]["result_policy"] = "consolidated-latest-attempt" + capability_count = len(selected_signatures) dataset_ids = sorted({str(row["dataset_id"]) for row in cases}) model_summaries: list[dict[str, Any]] = [] diff --git a/experiments/model-capability-benchmark/src/model_capability_bench/analytics/projector.py b/experiments/model-capability-benchmark/src/model_capability_bench/analytics/projector.py index 43b92a12..ddc65b82 100644 --- a/experiments/model-capability-benchmark/src/model_capability_bench/analytics/projector.py +++ b/experiments/model-capability-benchmark/src/model_capability_bench/analytics/projector.py @@ -410,6 +410,26 @@ def _create_schema(connection: Any) -> None: ) WHERE rn = 1; + CREATE OR REPLACE VIEW v_latest_quality_results AS + WITH consolidated AS ( + SELECT b.*, + ROW_NUMBER() OVER ( + PARTITION BY b.model_signature, b.capability_id, b.profile + ORDER BY b.completed_at_utc DESC, b.run_id DESC + ) AS rn + FROM benchmark_cells b + JOIN run_configurations rc ON rc.run_id = b.run_id + WHERE b.status = 'COMPLETED' AND rc.experiment_kind = 'consolidated' + ) + SELECT * EXCLUDE (rn) FROM consolidated WHERE rn = 1 + UNION ALL + SELECT q.* FROM v_current_quality_results q + WHERE NOT EXISTS ( + SELECT 1 FROM consolidated c + WHERE c.rn = 1 AND c.model_signature = q.model_signature + AND c.capability_id = q.capability_id AND c.profile = q.profile + ); + CREATE OR REPLACE VIEW v_model_history AS SELECT b.*, diff --git a/experiments/model-capability-benchmark/src/model_capability_bench/analytics/public_bundle.py b/experiments/model-capability-benchmark/src/model_capability_bench/analytics/public_bundle.py new file mode 100644 index 00000000..dc6ee1e2 --- /dev/null +++ b/experiments/model-capability-benchmark/src/model_capability_bench/analytics/public_bundle.py @@ -0,0 +1,80 @@ +"""Single-file public dashboard data, published with atomic replacement.""" +from __future__ import annotations + +import json +import os +import re +import tempfile +from datetime import UTC, datetime +from pathlib import Path +from typing import Any + +_PRIVATE_KEYS = { + "source_dir", "cwd", "executable", "model_path", "mmproj_path", "artifact_path", + "llama_server_bin", "local_path", "home", "hostname", "username", "user_name", + "environ", "environment_variables", "api_key", "authorization", "password", + "secret", "access_token", "refresh_token", "hf_token", "api_token", "auth_token", + "bearer_token", "token", +} +_LOCAL_PATH = re.compile(r"(?:/Users/|/home/|/private/|/var/|/tmp/|[A-Za-z]:[\\/])") + + +def public_value(value: Any) -> Any: + if isinstance(value, dict): + clean = {} + for key, item in value.items(): + name = key.lower() + if name in _PRIVATE_KEYS or name.endswith(("_api_key", "_password", "_secret")): + clean[key] = None + elif name.endswith("_json") and isinstance(item, str): + try: + decoded = json.loads(item) + except json.JSONDecodeError: + clean[key] = public_value(item) + else: + clean[key] = json.dumps(public_value(decoded), sort_keys=True) + else: + clean[key] = public_value(item) + return clean + if isinstance(value, list): + return [public_value(item) for item in value] + if isinstance(value, str) and ( + _LOCAL_PATH.search(value) or re.search(r"https?://[^/\s]+:[^/\s]+@", value) + ): + return "[redacted local or credential-bearing value]" + return value + + +def write_public_bundle(data_dir: Path, *, destination: Path) -> Path: + """Read only files referenced by this export's index, excluding stale files.""" + index = json.loads((data_dir / "index.json").read_text()) + + def read(relative: str) -> Any: + path = (data_dir / relative).resolve() + if not path.is_relative_to(data_dir.resolve()): + raise ValueError("Dashboard index references a file outside its export directory") + return json.loads(path.read_text()) + + bundle = public_value({ + "schema_version": "1", + "generated_at_utc": datetime.now(UTC).isoformat(), + "overview": read(index["overview"]), + **{section: {key: read(relative) for key, relative in index[section].items()} + for section in ("capabilities", "models", "runs")}, + }) + # Serialize first: failures must leave the previously published file intact. + content = json.dumps(bundle, ensure_ascii=False, sort_keys=True, separators=(",", ":")) + "\n" + destination = destination.resolve() + destination.parent.mkdir(parents=True, exist_ok=True) + fd, temp_name = tempfile.mkstemp(prefix=f".{destination.name}.", dir=destination.parent) + temporary = Path(temp_name) + try: + with os.fdopen(fd, "w", encoding="utf-8") as handle: + handle.write(content) + handle.flush() + os.fsync(handle.fileno()) + os.chmod(temporary, 0o644) + os.replace(temporary, destination) + finally: + temporary.unlink(missing_ok=True) + return destination diff --git a/experiments/model-capability-benchmark/src/model_capability_bench/cli.py b/experiments/model-capability-benchmark/src/model_capability_bench/cli.py index f190f8a8..b27eb82b 100644 --- a/experiments/model-capability-benchmark/src/model_capability_bench/cli.py +++ b/experiments/model-capability-benchmark/src/model_capability_bench/cli.py @@ -34,9 +34,11 @@ load_sweep, ) from model_capability_bench.runner.config import load_runner_defaults +from model_capability_bench.runner.consolidation import consolidate_runs from model_capability_bench.runner.manifest import write_run_artifacts from model_capability_bench.runner.planning import plan_benchmark from model_capability_bench.runner.progress import TerminalProgress +from model_capability_bench.runner.recovery import select_error_cases from model_capability_bench.runtimes import RegistryRuntimeResolver from model_capability_bench.sharing import create_share_snapshot, render_share_snapshot from model_capability_bench.suite import ( @@ -80,6 +82,12 @@ def build_parser() -> argparse.ArgumentParser: run.add_argument("--run-id") run.add_argument("--output-dir", type=Path) run.add_argument("--cache-dir", type=Path) + run.add_argument("--max-output-tokens", type=int) + run.add_argument( + "--rerun-errors-from", type=Path, action="append", default=[], + help="Select invalid/failed latest cases from a source run directory; repeat for multiple runs.", + ) + run.add_argument("--plan-only", action="store_true", help="Print selection without inference.") run.add_argument( "--progress", action=argparse.BooleanOptionalAction, @@ -158,6 +166,15 @@ def build_parser() -> argparse.ArgumentParser: project.add_argument("--rebuild", action="store_true") project.add_argument("--export-dashboard", action="store_true") + project.add_argument("--result-view", choices=("latest", "standard"), default="latest") + + consolidate = sub.add_parser("consolidate", help="Create an immutable latest-attempt consolidated run.") + consolidate.add_argument("--run-dir", type=Path, action="append", required=True) + consolidate.add_argument("--run-id", required=True) + consolidate.add_argument("--run-group", required=True) + consolidate.add_argument("--output-dir", type=Path) + consolidate.add_argument("--plan-only", action="store_true") + dashboard_data = sub.add_parser( "dashboard-data", help="Export typed dashboard JSON from the analytics database.", @@ -165,6 +182,7 @@ def build_parser() -> argparse.ArgumentParser: dashboard_data.add_argument("--results-root", type=Path) dashboard_data.add_argument("--database", type=Path) dashboard_data.add_argument("--output-dir", type=Path) + dashboard_data.add_argument("--result-view", choices=("latest", "standard"), default="latest") dashboard_build = sub.add_parser( "dashboard-build", @@ -262,6 +280,19 @@ def main() -> int: args = build_parser().parse_args() root = args.root.resolve() + if args.command == "consolidate": + output = args.output_dir or root / "results" / "runs" / args.run_id + summary = consolidate_runs( + sources=args.run_dir, output_dir=output, run_id=args.run_id, + run_group=args.run_group, suite=load_capability_suite(root), plan_only=args.plan_only, + ) + if not args.plan_only: + reporting_config = load_reporting_config(root) + report = load_benchmark_report(output, reporting_config) + write_report(report, reporting_config, output) + _json(summary) + return 0 + if args.command == "project": results_root = ( args.results_root.resolve() @@ -288,6 +319,8 @@ def main() -> int: payload["dashboard"] = export_dashboard_data( database_path=database, output_dir=results_root / "analytics" / "dashboard", + result_view=args.result_view, + public_output_path=results_root / "published" / "dashboard-data.json", ) _json(payload) return 0 @@ -312,6 +345,8 @@ def main() -> int: export_dashboard_data( database_path=database, output_dir=output_dir, + result_view=args.result_view, + public_output_path=results_root / "published" / "dashboard-data.json", ) ) return 0 @@ -615,7 +650,37 @@ def main() -> int: if args.cache_dir is not None else defaults.cache_dir ) - + if args.max_output_tokens is not None and args.max_output_tokens < 1: + raise ValueError("--max-output-tokens must be positive") + inference_config = ( + {"max_output_tokens": args.max_output_tokens} + if args.max_output_tokens is not None else {} + ) + selected_cases = None + recovery_metadata = {} + if args.rerun_errors_from: + if output_dir.resolve() in {path.resolve() for path in args.rerun_errors_from}: + raise ValueError("Recovery output must differ from source run directories") + selected_cases, recovery_metadata = select_error_cases( + args.rerun_errors_from, models=model_keys, capabilities=capability_ids, + profile=profile, seed=seed, + ) + if resolved_run_id in {item["run_id"] for item in recovery_metadata["recovery_sources"]}: + raise ValueError("Recovery requires a new run ID") + existing_manifest = output_dir / "run_manifest.json" + if existing_manifest.exists(): + existing = json.loads(existing_manifest.read_text()) + previous = existing["run"].get("metadata", {}) + if (existing["configuration"]["experiment_kind"] != "recovery" + or existing["configuration"]["inference"] != inference_config + or any(previous.get(key) != value for key, value in recovery_metadata.items())): + raise ValueError("Use a new run ID: recovery selection or configuration changed") + elif output_dir.exists() and any(output_dir.iterdir()): + raise ValueError("Recovery requires an empty output directory or a matching recovery manifest") + if args.plan_only: + _json({"selected_cases": selected_cases, "selected_case_count": len(selected_cases) if selected_cases else None, + "inference": inference_config, "output_dir": str(output_dir), "metadata": recovery_metadata}) + return 0 config = RunnerConfig( run_group=args.run_group, profile=profile, @@ -625,6 +690,9 @@ def main() -> int: resume=resume, retry_failures=retry_failures, run_id=resolved_run_id, + inference_config=inference_config, + selected_cases=selected_cases, + metadata=recovery_metadata, ) with TerminalProgress(enabled=args.progress) as progress: runtime_resolver = RegistryRuntimeResolver(os.environ) diff --git a/experiments/model-capability-benchmark/src/model_capability_bench/providers/openai_json.py b/experiments/model-capability-benchmark/src/model_capability_bench/providers/openai_json.py index 93ed0a9d..e02b0019 100644 --- a/experiments/model-capability-benchmark/src/model_capability_bench/providers/openai_json.py +++ b/experiments/model-capability-benchmark/src/model_capability_bench/providers/openai_json.py @@ -17,7 +17,6 @@ from benchmark_core.transports import create_openai_compatible_client from openai import OpenAI - from model_capability_bench.providers.parsing import parse_chat_response_content diff --git a/experiments/model-capability-benchmark/src/model_capability_bench/runner/consolidation.py b/experiments/model-capability-benchmark/src/model_capability_bench/runner/consolidation.py new file mode 100644 index 00000000..a387b17b --- /dev/null +++ b/experiments/model-capability-benchmark/src/model_capability_bench/runner/consolidation.py @@ -0,0 +1,230 @@ +"""Immutable, latest-attempt consolidation of compatible benchmark evidence.""" +from __future__ import annotations + +import copy +import hashlib +import json +import shutil +import tempfile +from collections import defaultdict +from datetime import UTC, datetime +from pathlib import Path +from typing import Any + +from benchmark_core import read_jsonl_records, sha256_file + +from model_capability_bench.runner.aggregation import aggregate_capability +from model_capability_bench.suite import CapabilitySuiteBundle + + +def _digest(value: Any) -> str: + return hashlib.sha256(json.dumps(value, sort_keys=True).encode()).hexdigest() + + +def consolidate_runs( + *, sources: list[Path], output_dir: Path, run_id: str, run_group: str, + suite: CapabilitySuiteBundle, plan_only: bool = False, +) -> dict[str, Any]: + """Select by logical identity and terminal timestamp; never select by score.""" + output_dir = output_dir.resolve() + if not run_id.strip() or not run_group.strip(): + raise ValueError("Consolidation needs non-empty run ID and group") + if output_dir.exists(): + raise ValueError("Consolidation output already exists; choose a new run ID") + source_paths = sorted({p.resolve() for p in sources}) + if not source_paths: + raise ValueError("Consolidation requires source runs") + selected: dict[tuple, dict] = {} + provenance = [] + model_defs: dict[str, dict] = {} + cap_defs: dict[str, dict] = {} + identities: dict[tuple, tuple] = {} + suite_identity = None + source_ids: set[str] = set() + for path in source_paths: + if path == output_dir or path in output_dir.parents: + raise ValueError("Consolidation output must be separate from its sources") + manifest = json.loads((path / "run_manifest.json").read_text()) + source_id = manifest["run"]["run_id"] + if source_id == run_id or source_id in source_ids: + raise ValueError("Source run IDs must be unique and differ from output run ID") + source_ids.add(source_id) + identity = manifest["suite"] + if suite_identity is not None and identity != suite_identity: + raise ValueError("Source suite versions, profiles and seeds must match") + suite_identity = identity + checksums = {name: sha256_file(path / name) for name in ( + "run_manifest.json", "state.jsonl", "raw.jsonl", "evaluation.jsonl", + )} + provenance.append({"run_id": source_id, "checksums": checksums}) + for model in manifest["models"]: + key = model["model_key"] + if key in model_defs and model_defs[key]["model_signature"] != model["model_signature"]: + raise ValueError("Cannot consolidate different artifacts under the same model key") + model_defs[key] = model + for cap in manifest["capabilities"]: + key = cap["capability_id"] + if key in cap_defs and any(cap_defs[key].get(k) != cap.get(k) for k in ( + "task_id", "dataset_ids", "metrics", + )): + raise ValueError("Source capability definitions must match") + cap_defs[key] = cap + raw = {(row["case_id"], row["attempt"]): row + for row in read_jsonl_records(path / "raw.jsonl")} + evaluations = {(row["case_id"], row["attempt"]): row + for row in read_jsonl_records(path / "evaluation.jsonl")} + case_identities = {} + for state in read_jsonl_records(path / "state.jsonl"): + cid = state["case_id"] + if "identity" in state: + case_identities[cid] = state["identity"] + case = case_identities.get(cid) + if case is None: + raise ValueError("Source state has no case identity") + fields = ( + "model_key", "capability_id", "dataset_id", "dataset_revision", + "dataset_split", "sample_id", "task_id", "task_version", + "prompt_id", "prompt_version", "suite_id", "suite_version", + ) + if any(not case.get(k) for k in fields): + raise ValueError("Source case identity is incomplete") + logical = tuple(case[k] for k in fields) + semantic_key = (case["model_key"], case["capability_id"], case["dataset_id"]) + semantic = tuple(case[k] for k in ( + "dataset_revision", "dataset_split", "task_version", "prompt_id", "prompt_version", + )) + if semantic_key in identities and identities[semantic_key] != semantic: + raise ValueError("Dataset revisions/splits and task/prompt versions must match") + identities[semantic_key] = semantic + rank = (datetime.fromisoformat(state["timestamp_utc"]), source_id, + int(state["attempt"]), cid) + if logical not in selected or rank > selected[logical]["rank"]: + selected[logical] = { + "rank": rank, "source_run_id": source_id, "state": state, + "identity": case, "raw": raw.get((cid, state["attempt"])), + "evaluation": evaluations.get((cid, state["attempt"])), + } + # A source changing while read cannot produce a reproducible snapshot. + if any(sha256_file(path / name) != value for name, value in checksums.items()): + raise ValueError("Source evidence changed during consolidation; stop the source run first") + if not selected: + raise ValueError("Source runs contain no cases") + evidence = [selected[key] for key in sorted(selected)] + for item in evidence: + if item["state"]["status"] != "completed" or not item["raw"] or not item["evaluation"]: + raise ValueError("Latest attempt is incomplete or lacks paired inference/evaluation evidence; recover it first") + current_caps = {cap.spec.capability_id: cap.spec for cap in suite.resolved_capabilities} + for key, cap in cap_defs.items(): + current = current_caps.get(key) + if current is None or [ + {"name": m.name, "source": m.source, "reducer": m.reducer, + "field": m.field, "primary": m.primary} for m in current.metrics + ] != cap["metrics"]: + raise ValueError("Current metric definitions differ from source; cannot recalculate safely") + generation = sorted({json.dumps(item["identity"]["generation"], sort_keys=True) for item in evidence}) + invalid = sum(not item["raw"]["record"]["valid"] or not item["evaluation"]["record"]["valid"] + for item in evidence) + summary = { + "run_id": run_id, "run_group": run_group, "planned_cases": len(evidence), + "completed_cases": len(evidence), "failed_cases": 0, "skipped_cases": 0, + "model_failures": 0, "output_dir": str(output_dir), + "invalid_cases": invalid, "failure_rate": invalid / len(evidence), + "generation_configs": [json.loads(g) for g in generation], + "mixed_generation": len(generation) > 1, + } + if plan_only: + return summary + now = datetime.now(UTC).isoformat() + signatures = {"models": {}, "executions": {}, "benchmarks": {}} + for key, model in model_defs.items(): + signatures["models"][key] = model["model_signature"] + executions = sorted({item["raw"]["metadata"]["execution_signature"] for item in evidence + if item["identity"]["model_key"] == key}) + execution = executions[0] if len(executions) == 1 else "sha256:execution:" + _digest(executions) + signatures["executions"][key] = execution + model_defs[key] = {**model, "execution_signature": execution, "execution_metadata": { + "runtime_source": "consolidated", "source_execution_signatures": executions, + }} + for cap in cap_defs: + cap_cases = [item for item in evidence if item["identity"]["capability_id"] == cap] + signatures["benchmarks"][cap] = "sha256:consolidated:" + _digest([ + {"identity": item["identity"], "source_run_id": item["source_run_id"], + "case_id": item["state"]["case_id"], "attempt": item["state"]["attempt"]} + for item in cap_cases + ]) + metadata = {"experiment_kind": "consolidated", "selection_policy": "latest-attempt-v1", + "sources": provenance, "signatures": signatures, + "mixed_generation": summary["mixed_generation"], + "generation_configs": summary["generation_configs"]} + groups = defaultdict(list) + for item in evidence: + case = item["identity"] + meta = {**item["state"]["metadata"], "run_id": run_id, "run_group": run_group, + "benchmark_signature": signatures["benchmarks"][case["capability_id"]], + "execution_signature": signatures["executions"][case["model_key"]], + "consolidation_source": { + "run_id": item["source_run_id"], "case_id": item["state"]["case_id"], + "attempt": item["state"]["attempt"], + "timestamp_utc": item["state"]["timestamp_utc"], + "generation": case["generation"], + "benchmark_signature": item["state"]["metadata"]["benchmark_signature"], + "execution_signature": item["state"]["metadata"]["execution_signature"], + }} + for kind in ("raw", "evaluation"): + row = copy.deepcopy(item[kind]) + row["metadata"] = meta + row["record"].update({"run_id": run_id}) + if "run_group" in row["record"]: + row["record"]["run_group"] = run_group + item[kind] = row + item["case_id"] = item["state"]["case_id"] + item["attempt"] = item["state"]["attempt"] + item["state"] = {**item["state"], "metadata": meta, "identity": case} + groups[(case["model_key"], case["capability_id"])].append(item) + aggregates = [] + indices = [] + for (model, cap), items in groups.items(): + records = aggregate_capability(current_caps[cap], items, model_key=model, + run_id=run_id, run_group=run_group, + profile=suite_identity["profile"]) + for record in records: + record.update({"model_signature": signatures["models"][model], + "execution_signature": signatures["executions"][model], + "benchmark_signature": signatures["benchmarks"][cap]}) + aggregates.extend(records) + indices.append({"run_id": run_id, "run_group": run_group, "model_key": model, + "capability_id": cap, "task_id": current_caps[cap].task_id, + "profile": suite_identity["profile"], "failure_count": 0, + "dataset_ids": list(current_caps[cap].dataset_ids), + "cases": [{"case_id": i["case_id"], "attempt": i["attempt"]} for i in items]}) + summary.update({"aggregate_count": len(aggregates), "metadata": metadata}) + manifest = { + "schema_version": "1", "created_at_utc": now, "run": summary, "suite": suite_identity, + "models": list(model_defs.values()), "capabilities": list(cap_defs.values()), + "signatures": signatures, "configuration": { + "configuration_id": "consolidated", "experiment_kind": "consolidated", + "inference": {"mixed_generation": summary["mixed_generation"], + "generation_configs": summary["generation_configs"]}, + }, + } + for cap in manifest["capabilities"]: + cap["benchmark_signature"] = signatures["benchmarks"][cap["capability_id"]] + output_dir.parent.mkdir(parents=True, exist_ok=True) + staging = Path(tempfile.mkdtemp(prefix=".consolidate-", dir=output_dir.parent)) + try: + def write_jsonl(name: str, rows: list) -> None: + (staging / name).write_text("".join(json.dumps(row, sort_keys=True) + "\n" for row in rows)) + for kind in ("state", "raw", "evaluation"): + write_jsonl(kind + ".jsonl", [item[kind] for item in evidence]) + write_jsonl("aggregates.jsonl", aggregates) + write_jsonl("report_index.jsonl", indices) + write_jsonl("events.jsonl", [{"event": "run_completed", "timestamp_utc": now, + "metadata": {"run_id": run_id}}]) + (staging / "run_manifest.json").write_text(json.dumps(manifest, indent=2, sort_keys=True)) + # No single-machine or single-budget performance claim for a consolidated run. + (staging / "environment.json").write_text(json.dumps({"consolidation": metadata})) + staging.rename(output_dir) + finally: + if staging.exists(): + shutil.rmtree(staging) + return summary diff --git a/experiments/model-capability-benchmark/src/model_capability_bench/runner/contracts.py b/experiments/model-capability-benchmark/src/model_capability_bench/runner/contracts.py index d4a2e3b4..f26f9f99 100644 --- a/experiments/model-capability-benchmark/src/model_capability_bench/runner/contracts.py +++ b/experiments/model-capability-benchmark/src/model_capability_bench/runner/contracts.py @@ -33,6 +33,7 @@ class RunnerConfig: inference_config: Mapping[str, object] = field(default_factory=dict) runtime_config: Mapping[str, object] = field(default_factory=dict) metadata: Mapping[str, object] = field(default_factory=dict) + selected_cases: tuple[tuple[str, str, str, str], ...] | None = None def __post_init__(self) -> None: if not self.run_group.strip(): @@ -49,6 +50,8 @@ def __post_init__(self) -> None: raise ValueError("capability_ids must not contain empty values") if not self.configuration_id.strip(): raise ValueError("configuration_id must not be empty") + if self.selected_cases is not None and not self.selected_cases: + raise ValueError("selected_cases must not be empty") @dataclass(frozen=True) diff --git a/experiments/model-capability-benchmark/src/model_capability_bench/runner/engine.py b/experiments/model-capability-benchmark/src/model_capability_bench/runner/engine.py index 10d45d49..f0cc7f8c 100644 --- a/experiments/model-capability-benchmark/src/model_capability_bench/runner/engine.py +++ b/experiments/model-capability-benchmark/src/model_capability_bench/runner/engine.py @@ -122,6 +122,12 @@ def run(self, config: RunnerConfig) -> RunnerSummary: model_keys=config.model_keys, capability_ids=config.capability_ids, ) + selected_cases = set(config.selected_cases) if config.selected_cases is not None else None + if selected_cases is not None: + matrix = tuple(arm for arm in matrix if any( + key[:3] == (arm.model_key, arm.capability_id, arm.dataset_id) + for key in selected_cases + )) capability_ids = tuple( dict.fromkeys(arm.capability_id for arm in matrix) ) @@ -331,7 +337,13 @@ def run(self, config: RunnerConfig) -> RunnerSummary: metadata={ **capability_metadata, "planned_cases": sum( - len(loaded_for_capability[dataset_id].samples) + sum( + selected_cases is None or ( + model.model.model_key, capability.spec.capability_id, + dataset_id, sample.sample_id, + ) in selected_cases + for sample in loaded_for_capability[dataset_id].samples + ) for dataset_id in capability.spec.dataset_ids ), }, @@ -340,7 +352,19 @@ def run(self, config: RunnerConfig) -> RunnerSummary: for dataset_id in capability.spec.dataset_ids: loaded = loaded_for_capability[dataset_id] - planned_cases += len(loaded.samples) + samples = tuple(sample for sample in loaded.samples if ( + selected_cases is None or ( + model.model.model_key, capability.spec.capability_id, + dataset_id, sample.sample_id, + ) in selected_cases + )) + if selected_cases is not None: + expected = {key[3] for key in selected_cases if key[:3] == ( + model.model.model_key, capability.spec.capability_id, dataset_id, + )} + if expected != {sample.sample_id for sample in samples}: + raise ValueError("Recovery samples no longer match the current dataset selection") + planned_cases += len(samples) task_context = TaskExecutionContext( run_id=identity.run_id, dataset_id=dataset_id, @@ -349,7 +373,7 @@ def run(self, config: RunnerConfig) -> RunnerSummary: metadata=context_metadata, ) - for sample in loaded.samples: + for sample in samples: case_identity = BenchmarkCaseIdentity( suite_id=self.suite.suite.suite_id, suite_version=self.suite.suite.version, @@ -602,7 +626,7 @@ def run(self, config: RunnerConfig) -> RunnerSummary: self.evidence_store.complete_case( case_id, attempt, - metadata=case_metadata, + metadata={**case_metadata, "valid": bool(task_result.valid)}, ) completed_cases += 1 self.evidence_store.record_stage_event( diff --git a/experiments/model-capability-benchmark/src/model_capability_bench/runner/recovery.py b/experiments/model-capability-benchmark/src/model_capability_bench/runner/recovery.py new file mode 100644 index 00000000..48a573f2 --- /dev/null +++ b/experiments/model-capability-benchmark/src/model_capability_bench/runner/recovery.py @@ -0,0 +1,50 @@ +"""Select unsuccessful latest logical cases without rewriting source evidence.""" +from __future__ import annotations + +import hashlib +import json +from pathlib import Path + +CaseSelection = tuple[str, str, str, str] + + +def select_error_cases( + sources: list[Path], *, models: tuple[str, ...], capabilities: tuple[str, ...], + profile: str, seed: int, +) -> tuple[tuple[CaseSelection, ...], dict[str, object]]: + latest: dict[CaseSelection, dict] = {} + provenance = [] + for source in sorted({path.resolve() for path in sources}): + manifest = json.loads((source / "run_manifest.json").read_text()) + suite = manifest["suite"] + if suite["profile"] != profile or suite["seed"] != seed: + raise ValueError("Recovery must use the source profile and seed") + state_bytes = (source / "state.jsonl").read_bytes() + provenance.append({ + "run_id": manifest["run"]["run_id"], + "state_sha256": hashlib.sha256(state_bytes).hexdigest(), + }) + for line in state_bytes.decode().splitlines(): + row = json.loads(line) + meta = row["metadata"] + key = tuple(str(meta.get(k) or "") for k in ( + "model_key", "capability_id", "dataset_id", "sample_id", + )) + if key[0] not in models or key[1] not in capabilities: + continue + if not all(key): + raise ValueError("Source case is missing recovery identity fields") + previous = latest.get(key) + if previous is None or row["timestamp_utc"] > previous["timestamp_utc"]: + latest[key] = row + selected = tuple(sorted(key for key, row in latest.items() if ( + row["status"] == "failed" + or (row["status"] == "completed" and row["metadata"].get("valid") is False) + ))) + if not selected: + raise ValueError("No unsuccessful cases match the selected models/capabilities") + return selected, { + "experiment_kind": "recovery", + "recovery_sources": provenance, + "recovery_cases": [list(key) for key in selected], + } diff --git a/experiments/model-capability-benchmark/src/model_capability_bench/runtimes/external.py b/experiments/model-capability-benchmark/src/model_capability_bench/runtimes/external.py index 005e4eb3..9aae28c2 100644 --- a/experiments/model-capability-benchmark/src/model_capability_bench/runtimes/external.py +++ b/experiments/model-capability-benchmark/src/model_capability_bench/runtimes/external.py @@ -29,4 +29,3 @@ def release(self, model: ResolvedModel) -> None: closer = getattr(provider, "close", None) or getattr(provider, "release", None) if callable(closer): closer() - return None diff --git a/experiments/model-capability-benchmark/src/model_capability_bench/suite.py b/experiments/model-capability-benchmark/src/model_capability_bench/suite.py index cfe36ee8..621c7b66 100644 --- a/experiments/model-capability-benchmark/src/model_capability_bench/suite.py +++ b/experiments/model-capability-benchmark/src/model_capability_bench/suite.py @@ -20,7 +20,6 @@ from model_capability_bench.datasets import build_dataset_registry from model_capability_bench.tasks import build_task_registry - CAPABILITY_ALIASES: Mapping[str, str] = { "mathematics": "mathematical-reasoning", "math": "mathematical-reasoning", diff --git a/experiments/model-capability-benchmark/suite.yaml b/experiments/model-capability-benchmark/suite.yaml index b7201a6a..50642af5 100644 --- a/experiments/model-capability-benchmark/suite.yaml +++ b/experiments/model-capability-benchmark/suite.yaml @@ -8,7 +8,7 @@ suite: generation: temperature: 0.0 - max_output_tokens: 256 + max_output_tokens: 2048 seed: 42 capabilities: diff --git a/experiments/model-capability-benchmark/tests/test_consolidation.py b/experiments/model-capability-benchmark/tests/test_consolidation.py new file mode 100644 index 00000000..43fdd824 --- /dev/null +++ b/experiments/model-capability-benchmark/tests/test_consolidation.py @@ -0,0 +1,160 @@ +from __future__ import annotations + +import json +from pathlib import Path + +import duckdb +import pytest +from benchmark_core import read_jsonl_records, sha256_file + +from model_capability_bench.analytics.dashboard_export import export_dashboard_data +from model_capability_bench.analytics.projector import project_results +from model_capability_bench.runner.consolidation import consolidate_runs +from model_capability_bench.suite import load_capability_suite + +ROOT = Path(__file__).resolve().parents[1] +MODEL = "lfm2.5-24b-a2b-ud-q3-k-xl" +CAP = "mathematical-reasoning" + + +def _source(base: Path, name: str, samples: list[tuple[str, bool]], *, tokens: int, day: int) -> Path: + from benchmark_core import to_jsonable + + cap = next(c.spec for c in load_capability_suite(ROOT).resolved_capabilities if c.spec.capability_id == CAP) + path = base / "runs" / name + path.mkdir(parents=True) + stamp = f"2026-10-{day:02d}T10:00:00+00:00" + manifest = { + "suite": {"suite_id": "capability-core", "version": "3", "profile": "core", "seed": 42}, + "run": {"run_id": name, "run_group": "test", "planned_cases": len(samples), + "completed_cases": len(samples), "failed_cases": 0, "metadata": {"signatures": { + "models": {MODEL: "model-signature"}, "executions": {MODEL: "execution-signature"}, + "benchmarks": {CAP: f"benchmark-{tokens}"}, + }}}, + "created_at_utc": stamp, + "models": [{"model_key": MODEL, "model_id": "vendor/model", "model_signature": "model-signature", + "execution_signature": "execution-signature", "deployment": "local"}], + "capabilities": [{"capability_id": CAP, "task_id": cap.task_id, "dataset_ids": list(cap.dataset_ids), + "metrics": [{k: v for k, v in to_jsonable(m).items() if k != "options"} + for m in cap.metrics]}], + } + states, raw, evaluations = [], [], [] + for sample, valid in samples: + cid = f"case-{sample}-{tokens}" + meta = {"model_key": MODEL, "capability_id": CAP, "dataset_id": "math-reasoning-controlled-v3", + "dataset_revision": "v3", "dataset_split": "controlled", "sample_id": sample, + "model_signature": "model-signature", "execution_signature": "execution-signature", + "benchmark_signature": f"benchmark-{tokens}", "run_id": name, + "run_group": "test", "valid": valid} + identity = {**meta, "task_id": CAP, "task_version": "1", "prompt_id": "math", "prompt_version": "1", + "suite_id": "capability-core", "suite_version": "3", "generation": {"max_output_tokens": tokens}} + state = {"case_id": cid, "attempt": 1, "timestamp_utc": stamp, + "status": "completed", "metadata": meta, "identity": identity} + states.append(state) + record = {"run_id": name, "sample_id": sample, "valid": valid, "model_id": "vendor/model", + "task_id": CAP, "input_tokens": 10, "output_tokens": tokens, "latency_ms": 100, + "normalized_output": {"answer": "1"}, "raw_output": {"answer": "1"}} + raw.append({**state, "record": record}) + evaluations.append({**state, "record": { + "run_id": name, "sample_id": sample, "task_id": CAP, "valid": valid, + "expected": "1", "prediction": "1" if valid else None, + "metrics": [{"name": "final_answer_accuracy", "value": int(valid), "primary": True}], + }}) + (path / "run_manifest.json").write_text(json.dumps(manifest)) + for filename, rows in (("state.jsonl", states), ("raw.jsonl", raw), ("evaluation.jsonl", evaluations)): + (path / filename).write_text("".join(json.dumps(row) + "\n" for row in rows)) + return path + + +def test_latest_attempt_merge_recalculates_metrics_and_preserves_sources(tmp_path: Path) -> None: + base = _source(tmp_path, "base", [("a", False), ("b", True), ("c", True)], tokens=2048, day=21) + retry = _source(tmp_path, "retry", [("a", True), ("b", False)], tokens=4096, day=22) + sources = [base, retry] + hashes = {p: sha256_file(p) for source in sources for p in source.iterdir()} + output = tmp_path / "runs" / "merged" + args = {"sources": sources, "output_dir": output, "run_id": "merged", "run_group": "test", + "suite": load_capability_suite(ROOT)} + preview = consolidate_runs(**args, plan_only=True) + assert not output.exists() + assert preview["planned_cases"] == 3 + assert preview["invalid_cases"] == 1 + summary = consolidate_runs(**args) + assert summary["mixed_generation"] is True + assert summary["failure_rate"] == pytest.approx(1 / 3) + raw = read_jsonl_records(output / "raw.jsonl") + evaluation = read_jsonl_records(output / "evaluation.jsonl") + assert len(raw) == len(evaluation) == 3 + assert {r["record"]["sample_id"]: r["record"]["valid"] for r in raw} == { + "a": True, "b": False, "c": True, + } + assert all(r["metadata"]["consolidation_source"]["run_id"] == "retry" for r in raw + if r["record"]["sample_id"] in {"a", "b"}) + primary = next(r for r in read_jsonl_records(output / "aggregates.jsonl") if r["primary"]) + assert primary["value"] == pytest.approx(2 / 3) + assert primary["sample_count"] == 3 + assert all(sha256_file(p) == digest for p, digest in hashes.items()) + # Source ordering has no effect on selected attempts or signatures. + other = tmp_path / "runs" / "other" + consolidate_runs(**{**args, "sources": list(reversed(sources)), "output_dir": other, "run_id": "other"}) + assert json.loads((other / "run_manifest.json").read_text())["signatures"] == json.loads( + (output / "run_manifest.json").read_text())["signatures"] + with pytest.raises(ValueError, match="already exists"): + consolidate_runs(**args) + + +def test_consolidated_projection_updates_latest_but_keeps_canonical_separate(tmp_path: Path) -> None: + source = _source(tmp_path, "source", [("a", True), ("b", False)], tokens=2048, day=21) + output = tmp_path / "runs" / "merged" + consolidate_runs(sources=[source], output_dir=output, run_id="merged", run_group="test", + suite=load_capability_suite(ROOT)) + projected = project_results(results_root=tmp_path, rebuild=True, run_dirs=(output,)) + assert projected.projected_runs == 1 + with duckdb.connect(projected.database_path, read_only=True) as db: + assert db.execute("SELECT COUNT(*) FROM v_current_quality_results").fetchone()[0] == 0 + assert db.execute("SELECT run_id FROM v_latest_quality_results").fetchall() == [("merged",)] + exported = export_dashboard_data(database_path=Path(projected.database_path), output_dir=tmp_path / "dashboard") + overview = json.loads(Path(exported["overview"]).read_text()) + public = json.loads(Path(exported["bundle"]).read_text()) + assert public["overview"]["decision"]["model_summaries"] == overview["decision"]["model_summaries"] + assert public["runs"]["merged"]["run"]["source_dir"] is None + model = overview["decision"]["model_summaries"][0] + assert model["observed_case_count"] == 2 + assert model["failure_rate"] == 0.5 + assert model["result_policy"] == "consolidated-latest-attempt" + assert overview["consolidations"][0]["run_id"] == "merged" + export_dashboard_data(database_path=Path(projected.database_path), output_dir=tmp_path / "standard", result_view="standard") + assert json.loads((tmp_path / "standard" / "overview.json").read_text())["cells"] == [] + + +def test_incomplete_latest_attempt_cannot_fall_back_to_success(tmp_path: Path) -> None: + source = _source(tmp_path, "source", [("a", True)], tokens=2048, day=21) + row = json.loads((source / "state.jsonl").read_text()) + row.update({"status": "failed", "timestamp_utc": "2026-10-22T10:00:00+00:00", "attempt": 2}) + with (source / "state.jsonl").open("a") as handle: + handle.write(json.dumps(row) + "\n") + with pytest.raises(ValueError, match="incomplete or lacks paired"): + consolidate_runs(sources=[source], output_dir=tmp_path / "merged", run_id="merged", + run_group="test", suite=load_capability_suite(ROOT)) + + +@pytest.mark.parametrize("change", ["profile", "artifact", "revision"]) +def test_incompatible_sources_are_rejected_before_writing(tmp_path: Path, change: str) -> None: + first = _source(tmp_path, "first", [("a", True)], tokens=2048, day=21) + second = _source(tmp_path, "second", [("a", False)], tokens=4096, day=22) + manifest_file = second / "run_manifest.json" + manifest = json.loads(manifest_file.read_text()) + if change == "profile": + manifest["suite"]["profile"] = "smoke" + elif change == "artifact": + manifest["models"][0]["model_signature"] = "different-artifact" + else: + state_file = second / "state.jsonl" + row = json.loads(state_file.read_text()) + row["identity"]["dataset_revision"] = "v4" + state_file.write_text(json.dumps(row) + "\n") + manifest_file.write_text(json.dumps(manifest)) + output = tmp_path / "merged" + with pytest.raises(ValueError): + consolidate_runs(sources=[first, second], output_dir=output, run_id="merged", + run_group="test", suite=load_capability_suite(ROOT)) + assert not output.exists() diff --git a/experiments/model-capability-benchmark/tests/test_limite_violetto.py b/experiments/model-capability-benchmark/tests/test_limite_violetto.py index bfb28dfe..b598df81 100644 --- a/experiments/model-capability-benchmark/tests/test_limite_violetto.py +++ b/experiments/model-capability-benchmark/tests/test_limite_violetto.py @@ -3,7 +3,6 @@ from pathlib import Path from types import SimpleNamespace -import pytest from benchmark_core import ( GenerationConfig, InferenceRequest, @@ -18,7 +17,7 @@ extract_last_boxed_answer, parse_chat_response_content, ) -from model_capability_bench.suite import CAPABILITY_ALIASES, resolve_capability_id +from model_capability_bench.suite import resolve_capability_id ROOT = Path(__file__).resolve().parents[1] REGISTRY = ROOT / "models.yaml" diff --git a/experiments/model-capability-benchmark/tests/test_projector.py b/experiments/model-capability-benchmark/tests/test_projector.py index c70f25e6..5389dea4 100644 --- a/experiments/model-capability-benchmark/tests/test_projector.py +++ b/experiments/model-capability-benchmark/tests/test_projector.py @@ -262,6 +262,25 @@ def _write_run( return run_dir +def test_recovery_cannot_replace_canonical_current_results(tmp_path: Path) -> None: + results = tmp_path / "results" + _write_run(results, run_id="canonical", completed_at="2026-10-05T08:00:00+00:00", score=0.7) + recovery = _write_run( + results, run_id="recovery", completed_at="2026-10-05T09:00:00+00:00", score=1.0, + ) + manifest_path = recovery / "run_manifest.json" + manifest = json.loads(manifest_path.read_text()) + manifest["configuration"] = {"experiment_kind": "recovery"} + manifest_path.write_text(json.dumps(manifest)) + summary = project_results(results_root=results, rebuild=True) + with duckdb.connect(summary.database_path, read_only=True) as connection: + for view in ("v_current_quality_results", "v_current_performance_results"): + assert connection.execute(f"SELECT run_id FROM {view}").fetchall() == [("canonical",)] + assert connection.execute("SELECT run_id FROM benchmark_cells ORDER BY run_id").fetchall() == [ + ("canonical",), ("recovery",), + ] + + def test_projector_selects_latest_completed_comparable_result(tmp_path: Path) -> None: results = tmp_path / "results" _write_run( diff --git a/experiments/model-capability-benchmark/tests/test_public_bundle.py b/experiments/model-capability-benchmark/tests/test_public_bundle.py new file mode 100644 index 00000000..e5d5af18 --- /dev/null +++ b/experiments/model-capability-benchmark/tests/test_public_bundle.py @@ -0,0 +1,73 @@ +from __future__ import annotations + +import json +from pathlib import Path + +import pytest + +from model_capability_bench.analytics.public_bundle import public_value, write_public_bundle + + +def _export(root: Path) -> None: + root.mkdir() + index = {"overview": "overview.json", "capabilities": {"cap": "cap.json"}, + "models": {"model": "model.json"}, "runs": {"run": "run.json"}} + (root / "index.json").write_text(json.dumps(index)) + for name in ("overview", "cap", "model", "run"): + (root / f"{name}.json").write_text(json.dumps({"name": name, "failure_rate": 0.03})) + (root / "stale.json").write_text(json.dumps({"stale": True})) + + +def test_public_bundle_contains_all_views_and_replaces_previous_snapshot(tmp_path: Path) -> None: + root = tmp_path / "export" + _export(root) + destination = tmp_path / "published" / "dashboard-data.json" + write_public_bundle(root, destination=destination) + data = json.loads(destination.read_text()) + assert data["schema_version"] == "1" + assert data["overview"]["name"] == "overview" + assert data["capabilities"]["cap"]["name"] == "cap" + assert data["models"]["model"]["name"] == "model" + assert data["runs"]["run"]["name"] == "run" + assert "stale" not in destination.read_text() + (root / "overview.json").write_text(json.dumps({"failure_rate": 0.01})) + write_public_bundle(root, destination=destination) + assert json.loads(destination.read_text())["overview"]["failure_rate"] == 0.01 + assert not list(destination.parent.glob(".dashboard-data.json.*")) + + +def test_failed_export_keeps_public_snapshot_intact(tmp_path: Path) -> None: + root = tmp_path / "export" + _export(root) + destination = tmp_path / "dashboard-data.json" + write_public_bundle(root, destination=destination) + before = destination.read_bytes() + (root / "cap.json").write_text("invalid JSON") + with pytest.raises(json.JSONDecodeError): + write_public_bundle(root, destination=destination) + assert destination.read_bytes() == before + + +def test_public_redaction_handles_serialized_metadata_and_keeps_metrics() -> None: + data = public_value({ + "source_dir": "/Users/private/runs", "failure_rate": 0.03, "output_tokens": 4096, + "execution_environment": {"cpu_model": "Apple M3 Pro", "total_memory_bytes": 100}, + "metadata_json": json.dumps({"model_path": "/Users/private/model.gguf", "api_key": "secret-value", "ctx_size": 8192}), + "error_message": "failure at /private/tmp/model.log", "url": "https://user:password@example.com", + }) + encoded = json.dumps(data) + assert "/Users/" not in encoded and "/private/" not in encoded + assert "secret-value" not in encoded and "user:password" not in encoded + assert data["failure_rate"] == 0.03 and data["output_tokens"] == 4096 + assert data["execution_environment"]["cpu_model"] == "Apple M3 Pro" + assert json.loads(data["metadata_json"])["ctx_size"] == 8192 + + +def test_bundle_rejects_index_path_traversal(tmp_path: Path) -> None: + root = tmp_path / "export" + _export(root) + index = json.loads((root / "index.json").read_text()) + index["overview"] = "../outside.json" + (root / "index.json").write_text(json.dumps(index)) + with pytest.raises(ValueError, match="outside"): + write_public_bundle(root, destination=tmp_path / "public.json") diff --git a/experiments/model-capability-benchmark/tests/test_recovery.py b/experiments/model-capability-benchmark/tests/test_recovery.py new file mode 100644 index 00000000..1f7c4f8e --- /dev/null +++ b/experiments/model-capability-benchmark/tests/test_recovery.py @@ -0,0 +1,77 @@ +from __future__ import annotations + +import json +import sys +from pathlib import Path + +import pytest + +from model_capability_bench.runner.recovery import select_error_cases + + +def _source(root: Path, rows: list[dict]) -> Path: + root.mkdir() + (root / "run_manifest.json").write_text(json.dumps({ + "suite": {"profile": "core", "seed": 42}, "run": {"run_id": root.name}, + })) + (root / "state.jsonl").write_text("\n".join(json.dumps(row) for row in rows)) + return root + + +def _row(sample: str, status: str, valid: bool, time: str = "1") -> dict: + return {"status": status, "timestamp_utc": time, "metadata": { + "model_key": "model", "capability_id": "cap", "dataset_id": "dataset", + "sample_id": sample, "valid": valid, + }} + + +def test_latest_logical_attempt_controls_recovery_across_sources(tmp_path: Path) -> None: + first = _source(tmp_path / "first", [ + _row("invalid", "completed", False), + _row("failed", "failed", False), + _row("fixed", "completed", False), + _row("valid", "completed", True), + _row("interrupted", "started", False), + ]) + second = _source(tmp_path / "second", [_row("fixed", "completed", True, "2")]) + before = (first / "state.jsonl").read_bytes() + selected, metadata = select_error_cases( + [second, first], models=("model",), capabilities=("cap",), profile="core", seed=42, + ) + assert {key[3] for key in selected} == {"invalid", "failed"} + assert metadata["experiment_kind"] == "recovery" + assert len(metadata["recovery_sources"]) == 2 + assert (first / "state.jsonl").read_bytes() == before + + +def test_recovery_rejects_different_profile_or_empty_selection(tmp_path: Path) -> None: + source = _source(tmp_path / "source", [_row("valid", "completed", True)]) + with pytest.raises(ValueError, match="profile and seed"): + select_error_cases([source], models=("model",), capabilities=("cap",), profile="smoke", seed=42) + with pytest.raises(ValueError, match="No unsuccessful cases"): + select_error_cases([source], models=("model",), capabilities=("cap",), profile="core", seed=42) + + +def test_cli_recovery_preview_and_source_protection(tmp_path: Path, monkeypatch, capsys) -> None: + from model_capability_bench.cli import main + + row = _row("bad", "completed", False) + row["metadata"]["model_key"] = "lfm2.5-24b-a2b-ud-q3-k-xl" + row["metadata"]["capability_id"] = "structured-output" + source = _source(tmp_path / "source", [row]) + output = tmp_path / "output" + args = [ + "model-bench", "run", "--models", row["metadata"]["model_key"], + "--capabilities", "structured-output", "--profile", "core", + "--run-group", "recovery", "--run-id", "recovery", "--output-dir", str(output), + "--max-output-tokens", "4096", "--rerun-errors-from", str(source), "--plan-only", + ] + monkeypatch.setattr(sys, "argv", args) + assert main() == 0 + payload = json.loads(capsys.readouterr().out) + assert payload["selected_case_count"] == 1 + assert payload["inference"]["max_output_tokens"] == 4096 + assert not output.exists() + args[args.index(str(output))] = str(source) + with pytest.raises(ValueError, match="must differ"): + main() diff --git a/experiments/model-capability-benchmark/tests/test_unified_runner.py b/experiments/model-capability-benchmark/tests/test_unified_runner.py index 76fb3770..2ea239cc 100644 --- a/experiments/model-capability-benchmark/tests/test_unified_runner.py +++ b/experiments/model-capability-benchmark/tests/test_unified_runner.py @@ -20,6 +20,51 @@ ROOT = Path(__file__).resolve().parents[1] +def test_recovery_executes_only_selected_samples_with_larger_budget(tmp_path: Path) -> None: + import json + + from model_capability_bench.runner.recovery import select_error_cases + + original_root = tmp_path / "original" + first = _runner(original_root, runtime=_FakeRuntime()) + config = RunnerConfig( + run_group="original", profile="smoke", model_keys=("qwen3.5-2b-q4km",), + capability_ids=("mathematical-reasoning",), + ) + first.run(config) + source = original_root / "evidence" + states = read_jsonl_records(source / "state.jsonl") + completed = [row for row in states if row["status"] == "completed"] + chosen = completed[0]["metadata"]["sample_id"] + for row in states: + if row["status"] == "completed" and row["metadata"]["sample_id"] == chosen: + row["metadata"]["valid"] = False + (source / "state.jsonl").write_text("\n".join(json.dumps(row) for row in states)) + (source / "run_manifest.json").write_text(json.dumps({ + "suite": {"profile": "smoke", "seed": 42}, "run": {"run_id": "original"}, + })) + selected, metadata = select_error_cases( + [source], models=config.model_keys, capabilities=config.capability_ids, + profile="smoke", seed=42, + ) + before = (source / "state.jsonl").read_bytes() + recovery = _runner(tmp_path / "recovery", runtime=_FakeRuntime()) + retry_config = dataclasses.replace( + config, run_group="recovery", selected_cases=selected, metadata=metadata, + inference_config={"max_output_tokens": 4096}, + ) + summary = recovery.run(retry_config) + assert summary.planned_cases == summary.completed_cases == 1 + raw = read_jsonl_records(tmp_path / "recovery" / "evidence" / "raw.jsonl") + assert [row["record"]["sample_id"] for row in raw] == [chosen] + started = read_jsonl_records(tmp_path / "recovery" / "evidence" / "state.jsonl")[0] + assert started["identity"]["generation"]["max_output_tokens"] == 4096 + assert (source / "state.jsonl").read_bytes() == before + resumed = recovery.run(retry_config) + assert resumed.completed_cases == 0 + assert resumed.skipped_cases == 1 + + class _FakeProvider: provider_id = "fake" diff --git a/korgis b/korgis index 3047842f..5d98b6e9 160000 --- a/korgis +++ b/korgis @@ -1 +1 @@ -Subproject commit 3047842f5c0ccba04d673588ea3b78db7ab7a7b4 +Subproject commit 5d98b6e951dfaba4a2a2583574b541cb8f09244c