chore: update model catalog from bot issues - #1094
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: bf9d4c7ab6
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".
| "grok-4.3": ["xAI", "openrouter", "bedrock"], | ||
| "grok-4.3-latest": ["xAI", "bedrock"], |
There was a problem hiding this comment.
When customers select the newly advertised Bedrock provider for grok-4.3 or grok-4.3-latest, this proxy still will not call Bedrock: both model specs remain format: "openai", and fetchModelLoop explicitly skips every secret.type === "bedrock" for OpenAI-format specs before routing, so Bedrock-only requests fall through to no usable key while mixed-key requests never exercise Bedrock. Please either leave Bedrock on a separately routable model id / actual gateway routing change, or add the missing format/routing support before listing it here.
AGENTS.md reference: AGENTS.md:L10-L12
Useful? React with 👍 / 👎.
The bot added `bedrock` to grok-4.3 (["xAI","openrouter"] -> +bedrock) and grok-4.3-latest (["xAI"] -> +bedrock), but both specs are `format: "openai"`, and the proxy deliberately skips bedrock secrets for OpenAI-format models (src/proxy.ts:1453 + src/proxy.bedrock-openai-skip.test.ts, #18324). So the added Bedrock provider is inert on these ids (confirms the codex P1 and AGENTS.md: routing changes belong in ../gateway/, and OpenAI-format ids must not advertise bedrock here). The Bedrock (Mantle) route for Grok 4.3 is already served by the dedicated `xai.grok-4.3` entry (format openai, available_providers ["bedrock"]) — validated 200 with real content via the :9090 gateway. So reverting these additions loses no coverage. Reverted grok-4.3 -> ["xAI","openrouter"] and grok-4.3-latest -> ["xAI"] in model_list.json + index.ts. Kept: the new accounts/fireworks/models/nemotron-lightning-3p5-30b-a3b (fireworks) — validated 200 both streaming and non-streaming via the gateway. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
1 participant
Automated daily batch of model catalog updates from bot issues.
Included issues
Summary
accounts/fireworks/models/nemotron-lightning-3p5-30b-a3baccounts/fireworks/models/nemotron-lightning-3p5-30b-a3b2
3
4
grok-4.3-latestgrok-4.3grok-4.3-latestgrok-4.32
3
Verified metadata
#1091: [BOT ISSUE] Add accounts/fireworks/models/nemotron-lightning-3p5-30b-a3b (Fireworks)
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.
#1093: [BOT ISSUE] Add bedrock to grok-4.3 available_providers (Bedrock)
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 no pricing/token discrepancies for the changed models.
No sync_models discrepancies were noted.