chore: update model catalog from bot issues#917
Conversation
github-actions
Bot
requested review from
Alex Z (CLowbrow),
aswink,
Caitlin Pinn (cpinn),
Erin McNulty (erin2722) and
Ken Jiang (knjiang)
|
Codex (@codex) review |
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
![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: 72db98a0f6
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".
| "displayName": "GPT-5", | ||
| "reasoning": true, | ||
| "max_input_tokens": 272000, | ||
| "max_input_tokens": 400000, |
There was a problem hiding this comment.
This field is the model's maximum input token limit, but the 400,000 value is the total context-window figure, not the accepted input size. OpenAI's GPT-5 developer announcement says the API models accept 272,000 input tokens plus 128,000 output/reasoning tokens for 400,000 total, and Azure's official GPT-5 table similarly lists the GPT-5 family as 400,000 context with Input: 272,000; this same 400,000 change is repeated for mini/nano/pro below, so catalog users will think prompts above the provider's input cap are valid and then hit upstream API errors. Sources: https://openai.com/index/introducing-gpt-5-for-developers/ and https://learn.microsoft.com/en-us/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure
Useful? React with 👍 / 👎.
…ndow #915/#916 changed gpt-5, gpt-5-mini, gpt-5-nano max_input_tokens 272000->400000 and gpt-5-pro 128000->400000, but 400000 is the total CONTEXT window, not the accepted input size. The catalog's max_input_tokens is the input cap: for the gpt-5 family max_input + max_output = 400000 (gpt-5/mini/nano: 272000 + 128000; gpt-5-pro: 128000 + 272000). OpenAI's model page confirms gpt-5 = 400k context / 128k output. Revert to the correct per-model input caps (Codex P1). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
…ues-2026-07-05 # Conflicts: # packages/proxy/schema/index.ts
1 participant
Automated daily batch of model catalog updates from bot issues.
Included issues
Summary
pearl-ai/gemma-4-31b-itpearl-ai/gemma-4-31b-it2
gpt-5gpt-5-minigpt-5-nanogpt-5gpt-5-minigpt-5-nano2
3
gpt-5-progpt-5-proVerified metadata
#914: [BOT ISSUE] Together: add missing pearl-ai/gemma-4-31b-it model
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.
#915: [BOT ISSUE] OpenAI: update stale max_input_tokens for gpt-5, gpt-5-mini, and gpt-5-nano
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.
#916: [BOT ISSUE] OpenAI: update stale max_input_tokens for gpt-5-pro
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.