chore: update model catalog from bot issues#989
Conversation
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
|
Codex (@codex) review |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 68bcbbe3fe
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "Codex (@codex) review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "Codex (@codex) address that feedback".
| "input_cost_per_mil_tokens": 0.5, | ||
| "output_cost_per_mil_tokens": 1.5, |
There was a problem hiding this comment.
These Vertex MaaS prices don't match Google's pricing for the same models: the official Agent Platform pricing page lists Llama 4 Maverick at $0.35 input / $1.15 output per MTok and Llama 4 Scout at $0.25 / $0.70, while this catalog sets Maverick to $0.50 / $1.50 and the Scout entry below to $0.20 / $0.60 (https://cloud.google.com/gemini-enterprise-agent-platform/generative-ai/pricing). Since these new catalog entries are provider-specific to Vertex, requests to either model will be mispriced.
Useful? React with 👍 / 👎.
| "locations": [ | ||
| "us-east5" | ||
| ], | ||
| "max_input_tokens": 1048576, |
There was a problem hiding this comment.
Google's current MaaS specs list llama-4-maverick-17b-128e-instruct-maas with max output 8,192 and context length 524,288, and Scout below with max output 8,192 and context length 1,310,720 (https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/llama/llama4-maverick and https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/llama/llama4-scout). Advertising 1,048,576 here and 10,485,760 for Scout, while omitting max_output_tokens, lets clients/UI that rely on this catalog accept prompts or output settings that Vertex will reject.
Useful? React with 👍 / 👎.
| "multimodal": true, | ||
| "input_cost_per_mil_tokens": 10, | ||
| "output_cost_per_mil_tokens": 50, | ||
| "input_cache_read_cost_per_mil_tokens": 1, |
There was a problem hiding this comment.
When a claude-mythos-5 request creates prompt-cache entries, the proxy emits cache_write_input_tokens, but this new catalog entry only supplies cache-hit/read pricing. Anthropic's pricing table lists Mythos 5 5-minute cache writes at $12.50/MTok and 1-hour writes at $20/MTok, matching Fable 5 (https://platform.claude.com/docs/en/about-claude/pricing), so cache-creation usage for this newly routed model will be unc costed or undercharged unless the input_cache_write_* fields are added.
Useful? React with 👍 / 👎.
1 participant
Automated daily batch of model catalog updates from bot issues.
Included issues
Summary
databricks-gpt-5-5-prodatabricks-gpt-5-4-minidatabricks-gemini-3-5-flashdatabricks-gpt-5-5-prodatabricks-gpt-5-4-minidatabricks-gemini-3-5-flash2
mistral-small-2603mistral-small-26032
grok-4.20-0309-non-reasoninggrok-4.20-0309-non-reasoning2
publishers/meta/models/llama-4-maverick-17b-128e-instruct-maaspublishers/meta/models/llama-4-scout-17b-16e-instruct-maaspublishers/meta/models/llama-4-maverick-17b-128e-instruct-maaspublishers/meta/models/llama-4-scout-17b-16e-instruct-maas2
3
4
5
claude-mythos-5claude-mythos-previewclaude-mythos-5claude-mythos-preview2
3
databricks-gpt-5-3-codexdatabricks-gemma-3-12bdatabricks-gpt-5-3-codexdatabricks-gemma-3-12b2
openai/gpt-oss-120bopenai/gpt-oss-120b2
Verified metadata
#975: [BOT ISSUE] Databricks: add missing databricks-gpt-5-5-pro, databricks-gpt-5-4-mini, databricks-gemini-3-5-flash
Verification notes
No LLM verification step ran; model metadata was already complete in the issue.
sync_models vs proposed update
sync_models cross-check found differences. Official provider verification was used for the applied values, and sync_models discrepancies are listed below for review.
#977: [BOT ISSUE] Mistral: add reasoning and reasoning_budget flags to mistral-small-2603
Verification notes
No LLM verification step ran; model metadata was already complete in the issue.
sync_models vs proposed update
sync_models cross-check found differences. Official provider verification was used for the applied values, and sync_models discrepancies are listed below for review.
#978: [BOT ISSUE] xAI: add missing input_cache_read_cost_per_mil_tokens to grok-4.20-0309-non-reasoning
Verification notes
No LLM verification step ran; model metadata was already complete in the issue.
sync_models vs proposed update
sync_models cross-check found differences. Official provider verification was used for the applied values, and sync_models discrepancies are listed below for review.
#980: [BOT ISSUE] Vertex: add missing Meta Llama 4 Maverick and Scout MaaS entries
Verification notes
Verification Checklist
sync_models vs proposed update
sync_models cross-check found differences. Official provider verification was used for the applied values, and sync_models discrepancies are listed below for review.
#982: [BOT ISSUE] Anthropic: add missing claude-mythos-5 and claude-mythos-preview models
Verification notes
No LLM verification step ran; model metadata was already complete in the issue.
sync_models vs proposed update
sync_models cross-check found differences. Official provider verification was used for the applied values, and sync_models discrepancies are listed below for review.
#987: [BOT ISSUE] Databricks: add missing databricks-gpt-5-3-codex and databricks-gemma-3-12b
Verification notes
No LLM verification step ran; model metadata was already complete in the issue.
sync_models vs proposed update
sync_models cross-check found differences. Official provider verification was used for the applied values, and sync_models discrepancies are listed below for review.
#988: [BOT ISSUE] Groq/Together: stale pricing for openai/gpt-oss-120b ($0.10/$0.50 → $0.15/$0.60)
Verification notes
Verification checklist
sync_models vs proposed update
sync_models cross-check found differences. Official provider verification was used for the applied values, and sync_models discrepancies are listed below for review.