chore: update model catalog from bot issues#1014
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: 9d14d43228
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".
| ] | ||
| }, | ||
| "google.gemma-3-12b-it": { | ||
| "format": "openai", |
There was a problem hiding this comment.
These new Bedrock-only entries use format: "openai" (the same pattern is added for google.gemma-3-4b-it and the openai.gpt-5.6-* Bedrock entries below), but this proxy never sends OpenAI-format requests with a Bedrock secret: packages/proxy/src/proxy.ts:1451-1455 skips secret.type === "bedrock" && modelSpec.format === "openai", and fetchOpenAI also throws for Bedrock. With only a Bedrock credential, requests for these newly advertised models fall through as No API keys found, so they should not be cataloged this way until the Bedrock OpenAI/Responses path is wired.
Useful? React with 👍 / 👎.
| "multimodal": true, | ||
| "input_cost_per_mil_tokens": 5.5, | ||
| "output_cost_per_mil_tokens": 33, | ||
| "input_cache_read_cost_per_mil_tokens": 0.55, |
There was a problem hiding this comment.
These GPT-5.6 Bedrock entries include cache-read pricing but omit input_cache_write_cost_per_mil_tokens; the AWS Bedrock pricing table lists 30-minute cache-write prices for Sol/Terra/Luna ($6.88/$3.44/$1.38 per 1M: https://aws.amazon.com/bedrock/pricing/). Any cost display or estimation that uses this catalog will undercharge cached-write tokens for these models.
Useful? React with 👍 / 👎.
| "flavor": "chat", | ||
| "displayName": "GLM 5.2", | ||
| "reasoning": true, | ||
| "max_input_tokens": 262144, |
There was a problem hiding this comment.
The Databricks supported-models page for endpoint databricks-glm-5-2 says GLM-5.2 supports a 1 million-token context length (https://docs.databricks.com/aws/en/machine-learning/foundation-model-apis/supported-models), but this catalog caps max_input_tokens at 262,144. Users and tooling that rely on the catalog will treat valid long-context GLM 5.2 prompts as over-limit or display the wrong capability.
Useful? React with 👍 / 👎.
1 participant
Automated daily batch of model catalog updates from bot issues.
Included issues
Summary
databricks-gemini-3-6-flashdatabricks-glm-5-2databricks-inklingdatabricks-gemini-3-6-flashdatabricks-glm-5-2databricks-inkling2
google.gemma-3-12b-itgoogle.gemma-3-4b-itnvidia.nemotron-nano-9b-v2google.gemma-3-12b-itgoogle.gemma-3-4b-itnvidia.nemotron-nano-9b-v22
3
4
openai.gpt-5.6-solopenai.gpt-5.6-terraopenai.gpt-5.6-lunaopenai.gpt-5.6-solopenai.gpt-5.6-terraopenai.gpt-5.6-luna2
3
4
5
Verified metadata
#1011: [BOT ISSUE] Databricks: add missing databricks-gemini-3-6-flash, databricks-glm-5-2, and databricks-inkling
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.
#1012: [BOT ISSUE] Bedrock: add missing google.gemma-3-12b-it, google.gemma-3-4b-it, and nvidia.nemotron-nano-9b-v2
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.
#1013: [BOT ISSUE] Add missing Bedrock entries for OpenAI GPT-5.6 (Sol, Terra, Luna)
Verification notes
Verification
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.