feat(fastwire): let a relay declare its service_tier echo non-authoritative

docs-site/src/content/docs/guides/codex-integration.md

@@ -131,6 +131,14 @@ echo `default`, so request logs show the response tier as an observation with co
 `assumed`. For latency-sensitive work, compare observed first-output times across the providers you
 actually use rather than assuming any particular channel is faster.
 
+For a gateway forwarding to a backend with the same metadata limitation, explicitly declare
+[`responseTierAuthoritative: false`](/reference/configuration/providers/#response-service-tier-authority)
+on that provider. The request still sends priority and records the raw echo, while actual Fast
+scheduling remains unconfirmed. Undeclared gateways and the official API keep their existing
+response-based interpretation. Updating OpenCodex alone does not add this declaration to existing
+gateway entries: without it, an eligible priority request followed by a `default` echo still records
+`response-declined`.
+
 The proxy listens on port `10100` by default and serves `POST /v1/responses`,
 `POST /v1/responses/compact`, `POST /v1/images/generations`, `POST /v1/images/edits`,
 `GET /v1/models`, `GET /healthz`, and the `/api/*` management surface.

docs-site/src/content/docs/reference/configuration/providers.md

@@ -205,6 +205,7 @@ Providers can expose a built-in shorthand, such as `agy` for `google-antigravity
 | `allowEncryptedV2AgentTasks?` | `boolean` | Disabled by default. Trust a direct key-auth `openai-responses` provider to consume or relay opaque encrypted V2 sub-agent tasks unchanged. Eligible routes skip `agentTaskRecovery`; all other routes keep the existing recovery or fail-closed behavior. OpenCodex does not decrypt, translate, or recover tasks sent through this opt-in. |
 | `upstreamWebsocket?` | `boolean` | Opt-in upstream Responses WebSocket transport for `openai-responses` requests (default false). Honored only for the first-party `https://api.openai.com/v1` upstream; custom-provider endpoints always use bounded HTTP/SSE because Bun cannot enforce an inbound WebSocket message limit before allocating the complete message. For the canonical ChatGPT `openai` provider, omitting it keeps the upstream WebSocket on eligible turns, an explicit `false` sends streaming turns over HTTP/SSE, and provider management rejects `true`; with `false`, native mid-turn steering and injection are unavailable. The field is independent of the client-facing `websockets` setting and changes neither the endpoint nor the credential. Plain HTTP remains on SSE; non-Responses paths and `openai-chat` requests stay on HTTP. |
 | `supportsServiceTier?` | `boolean` | Tri-state canonical Fast capability fallback. `true` publishes Fast in the catalog, satisfies service-tier routing requirements, contributes a supported fingerprint, and lets fast mode inject the provider's canonical wire value on a compatible final adapter. `false` strips the field and never injects, and exact model declarations cannot reopen it. Absent leaves the provider unclassified: fast mode does not inject or normalize a canonical caller value, and caller values obey the final wire's forwarding permission (`chatServiceTier` on Chat; passthrough on Responses). The registry classifies canonical OpenAI (`true`), DeepSeek, and Volcengine Ark (`false`); set it explicitly only for custom gateways that genuinely support tiers. |
+| `responseTierAuthoritative?` | `boolean` | Whether the response tier can confirm or deny Fast. Set `false` explicitly for routes with non-authoritative response metadata; omission preserves existing behavior. This does not enable Fast or change outbound parameters. See [Response service-tier authority](#response-service-tier-authority). |
 | `fastEnabled?` | `boolean` | Operator switch for the provider's Fast lane. `false` turns Fast off (no Fast toggle, no `--fast` row, no fast wire field) and overrides `supportsServiceTier`. `true` enables a lane the registry marks opt-in. Absent keeps the registry default: off for `anthropic` and `anthropic-apikey`, whose fast mode spends usage credits at 2x price, and unchanged for every other provider. The dashboard Models page shows an Off/On row for opt-in providers. |
 | `modelSupportsServiceTier?` | `Record<string, boolean>` | Exact upstream model capability overrides. Exact `true` enables canonical Fast for that model; exact `false` narrows provider defaults. An explicit provider-level `supportsServiceTier: false` remains fail-closed and cannot be reopened. Exact `true` does not authorize foreign caller-tier forwarding on Chat. Undeclared models fall back to provider-wide behavior. Management `PATCH /api/providers` merges entries and accepts `null` to clear one. |
 | `chatServiceTier?` | `boolean` | Provider-wide Chat-wire opt-in for forwarding caller `service_tier` values. On a classified route it governs foreign values such as `flex`, not proxy-owned canonical Fast after capability validation; on an unclassified route it governs every caller value because no Fast capability has been validated. Exact model capability does not authorize foreign forwarding. Responses routes retain their capability-based caller forwarding behavior. |
@@ -565,6 +566,44 @@ cleared or unresolved. The persisted catalog field is read by Codex for the curr
 model, which is why a valid configured selector is copied to each applicable entry.
 Provider-scoped selectors (above) are applied before this root fallback and win on routed rows.
 
+### Response service-tier authority
+
+Set `providers.<name>.responseTierAuthoritative` to `false` in `config.json` when that
+provider's response `service_tier` cannot establish whether Fast was granted. This is an
+operator declaration about the entire provider route, independent of `supportsServiceTier` and
+`fastWire`. It neither enables Fast nor changes the outgoing `service_tier`.
+
+For example, add this field to an existing gateway provider whose Codex backend has
+non-authoritative response metadata:
+
+```json
+{ "responseTierAuthoritative": false }
+```
+
+**Gateway support is opt-in.** Updating OpenCodex alone does not add this declaration to
+existing providers. Without it, an eligible priority request followed by `service_tier: "default"`
+retains the legacy `response-declined` interpretation. Configure `false` only when the
+route's response metadata is known to be non-authoritative.
+
+For an eligible request serialized as `priority`, both a `default` and a `priority` echo remain
+observations. Logs preserve `responseServiceTier` and record
+`tierOutcome.responseTierAuthoritative: false`, `fastOutcome: "applied"`, and
+`confirmation: "assumed"`, without `response-declined`. Here **applied means the request parameter
+was sent**, and **assumed means the actual Fast effect is unconfirmed**. The model tooltip shows
+the request and raw response separately with that confirmation. Neither this setting nor these
+records prove an acceleration or a billed tier.
+
+Cost estimates use the existing requested-tier fallback instead of treating the raw echo as a
+confirmed price tier; pricing rules requiring response confirmation cannot use that echo. Historical records
+without an authority flag retain their previous interpretation.
+
+The field accepts only booleans. Omission and explicit `true` retain response-based confirmation
+for other destinations, including the official API and undeclared gateways. Canonical
+`https://chatgpt.com/backend-api/codex` with `authMode: "forward"` remains automatically
+non-authoritative, even if `true` is configured. Gateway names and URLs are never inferred.
+If one gateway mixes response contracts, use separate provider entries for those routes and
+apply the declaration only to the relevant entry.
+
 ### FastWire B1 capability migration
 
 Fast capability and arbitrary Chat caller-tier forwarding are independent after FastWire B1. The

docs-site/src/content/docs/zh-cn/reference/configuration/providers.md

@@ -85,6 +85,7 @@ selector，而不是分配一个新名称。
 | `chatCompletionsPath?` | `string` | 用于 `openai-chat` 请求的相对资源路径，是 `responsesPath` 的对应项，适用相同的路径规则。当同一上游以不同前缀提供 Chat Completions 和 Responses 时需要此配置：按模型的 wire override 只更换适配器而不改动 `baseUrl`，否则已启用的 Chat 请求会被发送到 Responses base。随附示例为 Z.AI。 |
 | `upstreamWebsocket?` | `boolean` | 为 `openai-responses` 请求选择性启用上游 Responses WebSocket 传输（默认 `false`）。仅对第一方 `https://api.openai.com/v1` 上游生效；自定义提供者端点始终使用有界 HTTP/SSE，因为 Bun 无法在分配完整消息之前对入站 WebSocket 消息实施大小限制。对于规范 ChatGPT `openai` 提供商，省略该字段会在符合条件的轮次使用上游 WebSocket，`false` 通过 HTTP/SSE 发送流式轮次，`true` 会被拒绝；设为 `false` 时，原生轮次中操控与注入不可用。该字段独立于客户端侧的 `websockets` 设置，且不改变端点或凭据。普通 HTTP 仍使用 SSE；非 Responses 路径和 `openai-chat` 请求仍使用 HTTP。 |
 | `supportsServiceTier?` | `boolean` | `service_tier` 能力的三态。`true`：fast 模式可以注入，调用方提供的值也会被保留。`false`：剥离该字段且绝不注入（已明确不支持的上游不会收到它）。未设置：未分类——调用方提供的值原样保留，fast 模式绝不注入。注册表已对官方 OpenAI（`true`）、DeepSeek 和 Volcengine Ark（`false`）分类；仅对真正支持分层的自定义网关显式设置。 |
+| `responseTierAuthoritative?` | `boolean` | 响应等级是否足以确认或否定 Fast。对响应元数据不具权威性的链路显式设为 `false`；未设置时保持原有行为。不会启用 Fast 或修改出站参数。详见[响应服务等级的可信度](#响应服务等级的可信度)。 |
 | `preserveResponsesReasoningContent?` | `boolean` | 在重放的 Responses reasoning 项中保留明文 reasoning 内容，而不是清空（清空是 ChatGPT 后端的规则）。对接受 reasoning 重放的上游（如 DeepSeek）启用。代理生成的 `ocxr1` 信封始终会被剥离。 |
 | `disabled?` | `boolean` | 将提供者保留在磁盘上，但从路由和模型/目录列表中排除。 |
 | `apiKey?` | `string` | API key，或在请求时解析的 `${ENV_VAR}` / `$ENV_VAR` 引用。 |
@@ -172,6 +173,37 @@ API key 提供者可以持有字面量 key，或环境引用。OAuth 提供者
 
 `PATCH /api/providers?name=<provider>` 只修改它指定的字段，无论目的地如何都保留其他所有已存储字段。它接受全部五项设置，`null` 表示清除。对于两个推理列表，空数组会作为显式退出选项保存，而不会被删除。
 
+### 响应服务等级的可信度
+
+当提供者响应中的 `service_tier` 无法确定 Fast 是否生效时，在 `config.json` 中将
+`providers.<name>.responseTierAuthoritative` 设为 `false`。这是运营者对整个提供者链路的声明，
+独立于 `supportsServiceTier` 和 `fastWire`；它不会启用 Fast，也不会改变出站 `service_tier`。
+
+例如，对于已知 Codex 后端响应元数据不具权威性的网关，在现有 provider 对象中加入：
+
+```json
+{ "responseTierAuthoritative": false }
+```
+
+**网关需要显式配置。** 仅升级 OpenCodex 不会为现有 provider 自动添加该声明。未配置时，
+符合 Fast 条件且已发送 priority 的请求收到 `service_tier: "default"`，仍会按原逻辑记录为 `response-declined`。
+仅对已知响应元数据无法确认实际等级的链路设置 `false`。
+
+对于符合 Fast 条件且已发送 `priority` 的请求，`default` 和 `priority` 响应都只作为观察值。
+日志保留 `responseServiceTier`，并记录 `tierOutcome.responseTierAuthoritative: false`、
+`fastOutcome: "applied"` 和 `confirmation: "assumed"`，不会仅凭响应回显判为 `response-declined`。
+这里 **applied 表示请求参数已发出**，**assumed 表示 Fast 的实际效果尚未确认**。
+模型提示会分别显示请求等级、原始响应等级和确认状态。这些记录不能证明实际加速或计费等级。
+
+费用估算沿用请求等级的回退逻辑，不将原始回显当成已确认的价格等级；要求响应确认的定价规则
+不能使用该回显作为证据。没有可信度标记的历史记录保留原解释。
+
+该字段只接受布尔值。对于官方 API、未声明的网关等其他目的地，省略或显式设为 `true`
+都保留基于响应的判定。标准 `https://chatgpt.com/backend-api/codex` 地址配合
+`authMode: "forward"` 始终自动按非权威处理，即使配置为 `true` 也不例外。
+OpenCodex 不会依据网关名称或 URL 猜测可信度。如果同一网关混合不同的响应契约，
+应拆成独立 provider 条目，只为相关条目配置此声明。
+
 ## 提供者诊断出站安全性
 
 仪表板连接测试和实时模型发现使用受限的、仅 GET 传输。没有出站代理时，opencodex 只会解析一次主机名，并仅连接到该已验证地址。HTTPS 仍会保留原始 Host、SNI 和证书验证；提供者配置不能关闭证书检查。

scripts/test-layout/layout.json

@@ -966,6 +966,7 @@
     "fastwire-characterization-routing.test.ts": "routing",
     "fastwire-characterization-wire.test.ts": "routing",
     "fastwire-observability.test.ts": "routing",
+    "fastwire-response-authority.test.ts": "routing",
     "fastwire-policy.test.ts": "routing",
     "featherless-provider.test.ts": "providers",
     "fetch-header-timeout.test.ts": "server",

src/config/schema/leaf-validators.ts

@@ -297,6 +297,7 @@ export const providerConfigSchema = z.object({
   annotateEmptyToolOutputs: z.boolean().optional(),
   foldDeveloperRoleToSystem: z.boolean().optional(),
   fastWire: fastWireSchema.nullable().optional(),
+  responseTierAuthoritative: z.boolean().optional(),
   fastEnabled: z.boolean().optional(),
   supportsServiceTier: z.boolean().optional(),
   modelSupportsServiceTier: z.record(z.string().min(1), z.boolean()).optional(),

src/providers/fastwire.ts

@@ -340,6 +340,9 @@ export function createAdapterTierMetadata(
   const loggedWireValue = wireValue === null ? null : sanitizeLogMetadataString(wireValue);
   const outcome: AttemptTierOutcome = {
     wireKind,
+    ...(context.responseTierAuthoritative !== undefined
+      ? { responseTierAuthoritative: context.responseTierAuthoritative }
+      : {}),
     ...(wireValue === null
       ? { wireValue: null }
       : loggedWireValue ? { wireValue: loggedWireValue } : {}),

src/providers/model-rename-fields.ts

@@ -16,6 +16,7 @@ export const PROVIDER_MODEL_RENAME_ROLES = {
   mcpMaxResultBytes: "none",
   modelAdapters: "record",
   fastWire: "none",
+  responseTierAuthoritative: "none",
   baseUrl: "none",
   responsesPath: "none",
   chatCompletionsPath: "none",

src/providers/openai-tiers-destination.ts

@@ -25,6 +25,15 @@ export function isCanonicalOpenAiForwardProvider(provider: OcxProviderConfig): b
     && normalizedBaseUrl(provider.baseUrl) === CODEX_FORWARD_BASE_URL;
 }
 
+/**
+ * Response evidence is separate from Fast capability and request serialization. Preserve the
+ * known Codex exception (#2558); other destinations may declare their response contract without
+ * a gateway-name or URL heuristic. Absence retains the authoritative legacy default.
+ */
+export function responseTierAuthorityForProvider(provider: OcxProviderConfig): boolean | undefined {
+  return isCanonicalOpenAiForwardProvider(provider) ? false : provider.responseTierAuthoritative;
+}
+
 const OPENAI_API_ORIGIN = "https://api.openai.com";
 const OPENAI_API_BASE_URL = `${OPENAI_API_ORIGIN}/v1`;
 const OPENAI_API_RESPONSES_URL = `${OPENAI_API_BASE_URL}/responses`;

src/server/auth-cors.ts

@@ -973,6 +973,7 @@ const PROVIDER_CONFIG_FIELD_POLICY = {
   mcpMaxResultBytes: "editor",
   modelAdapters: "editor",
   fastWire: "editor",
+  responseTierAuthoritative: "editor",
   fastEnabled: "editor",
   baseUrl: "editor",
   responsesPath: "editor",

src/server/responses/core-normalize.ts

@@ -17,6 +17,7 @@ import { resolveOpenCodeGoTransport } from "../../providers/opencode-go-transpor
 import { getOrAllocateRequestSessionLane } from "../request-log-conversation";
 import { shouldPreparePlaintextV2AgentMessages } from "../../responses/plaintext-v2-agent-messages";
 import { hasValidatedActiveReasoningEffort } from "../../responses/parser";
+import { responseTierAuthorityForProvider } from "../../providers/openai-tiers-destination";
 import { isCanonicalOpenAiForwardProvider } from "../../providers/openai-tiers";
 import { applyOpenAiVirtualModel } from "../../providers/openai-virtual-models";
 import { renameRoutedIdentityInContext } from "../../adapters/identity";
@@ -220,14 +221,12 @@ export async function applyFinalRouteRequestNormalization(args: {
   );
   const modelServiceTierSupport = serviceTierSupportFromPolicy(fastPolicy);
   const callerTier = parsed.options.serviceTier;
-  // The ChatGPT-internal Codex backend echoes `service_tier: "default"` even on turns it
-  // scheduled as priority, so its echo cannot confirm OR deny Fast. Believing it reported every
-  // Fast request as `response-declined` (#2558). The public API's echo stays authoritative.
+  // Capture destination evidence policy separately from the unchanged outbound Fast decision.
   parsed.options.tierObservation = tierObservationContext(
     fastPolicy,
     config.fastMode,
     callerTier,
-    isCanonicalOpenAiForwardProvider(route.provider) ? false : undefined,
+    responseTierAuthorityForProvider(route.provider),
   );
   parsed.options.tierDecision = decideTier(fastPolicy, config.fastMode, callerTier);
   parsed.options.serviceTier = tierValueAfterDecision(parsed.options.tierDecision, callerTier);

src/types/provider.ts

@@ -218,6 +218,8 @@ export interface AttemptTierOutcome {
   callerFastSuppressedByConfig?: boolean;
   confirmation: "confirmed" | "assumed" | "downgraded" | "unknown";
   responseServiceTier?: string;
+  /** False retains the raw echo as evidence only, including during cost estimation. */
+  responseTierAuthoritative?: boolean;
 }
 
 /**
@@ -307,6 +309,13 @@ export interface OcxProviderConfig {
    * absence derives from the final model adapter.
    */
   fastWire?: FastWire | null;
+  /**
+   * Whether echoed service_tier can confirm or deny Fast. Set false for a relay whose
+   * response metadata cannot establish the granted tier. Absence keeps legacy authority;
+   * canonical ChatGPT Codex forwarding always treats the echo as non-authoritative.
+   * Observation only: this does not enable Fast or change request serialization.
+   */
+  responseTierAuthoritative?: boolean;
   baseUrl: string;
   /**
    * Optional relative resource path for key-auth openai-responses requests. Must start with `/`

src/usage/cost.ts

@@ -439,13 +439,16 @@ export function serviceTierContext(entry: ServiceTierContext): ServiceTierContex
 
 /** Convert one adapter-observed attempt outcome into the existing pricing provenance shape. */
 export function serviceTierContextFromOutcome(outcome: AttemptTierOutcome): ServiceTierContext {
-  if (outcome.canonical === "priority" && outcome.confirmation === "confirmed") {
+  const responseAuthoritative = outcome.responseTierAuthoritative !== false;
+  if (responseAuthoritative && outcome.canonical === "priority" && outcome.confirmation === "confirmed") {
     return { responseServiceTier: "priority" };
   }
-  if (outcome.responseServiceTier !== undefined) {
+  // Non-authoritative echoes remain in the log, but cannot become pricing confirmation either.
+  if (responseAuthoritative && outcome.responseServiceTier !== undefined) {
     return { responseServiceTier: outcome.responseServiceTier };
   }
-  if (outcome.canonical === "priority" && outcome.confirmation === "assumed") {
+  if (outcome.canonical === "priority"
+    && (outcome.confirmation === "assumed" || (!responseAuthoritative && outcome.confirmation === "confirmed"))) {
     return { requestedServiceTier: "priority" };
   }
   // An unclassified route makes no canonical Fast claim, but its adapter can still prove that