chore: sync new models - #1168
Conversation
Qwen2.5 VL 72B Instruct's input dropped to $0.25/M but its input_cache_read_cost_per_mil_tokens stayed at the previous tariff's $0.40/M — making cached input cost more than ordinary input. OpenRouter no longer publishes a cache-read rate for this model, so remove the stale field rather than invent one (addresses codex P2). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
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: 408fbe9193
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".
| "qwen/qwen3.8-flash": ["openrouter"], | ||
| "z-ai/glm-5.3-flash": ["openrouter"], |
There was a problem hiding this comment.
These entries modify AvailableEndpointTypes in the deprecated proxy even though both new records already declare openrouter through available_providers in model_list.json. Repository policy allows catalog changes here but requires routing changes to be made in the parent gateway, so remove these legacy mappings and apply any necessary routing update in ../gateway/ instead.
AGENTS.md reference: AGENTS.md:L10-L15
Useful? React with 👍 / 👎.
- qwen/qwen3.8-27b: union groq onto the existing (openrouter) entry and apply Groq's published metadata (input $0.8, output $4, cache-read $0.4, 131042 ctx, 16384 max output, multimodal, reasoning), matching the qwen3.6-27b groq+openrouter precedent (groq listed first). index.ts -> ["groq","openrouter"]. - zai-org/GLM-5.3-Flash: add the Baseten-served entry (baseten keys GLM models by their zai-org/ HF id, separate from the openrouter z-ai/ slug, same as GLM-5.2). Values from Baseten's /v1/models: input $0.15, output $0.5, cache-read $0.03, 1048576 ctx, 131072 max output, multimodal, reasoning. index.ts -> ["baseten"]. Both invoke 200 (streaming + non-streaming) through the gateway with real content. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
…data/braintrust-proxy into chore/sync-new-models # Conflicts: # packages/proxy/schema/model_list.json
1 participant
Automated sync of model metadata and newly available models from the approved provider allowlist.