chore: sync new models by github-actions[bot] · Pull Request #834 · braintrustdata/braintrust-proxy

github-actions · 2026-06-19T12:27:39Z

Automated sync of model metadata and newly available models from the approved provider allowlist.

vercel · 2026-06-19T12:27:43Z

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
ai-proxy	Ready	Preview, Comment	Jun 22, 2026 1:14pm

github-actions · 2026-06-19T12:27:46Z

Codex (@codex) review

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 40df2b1d36

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".

chatgpt-codex-connector · 2026-06-19T12:30:56Z

+  "openai.gpt-5.5": ["bedrock"],
+  "openai.gpt-5.4": ["bedrock"],
  "nvidia/nemotron-3-ultra-550b-a55b": ["together"],
  "moonshot.kimi-k2-thinking": ["bedrock"],
+  "xai.grok-4.3": ["bedrock"],


Use a Bedrock-compatible format for these endpoints

These new overrides make getModelEndpointTypes() route openai.gpt-5.5, openai.gpt-5.4, and xai.grok-4.3 to Bedrock, but their catalog specs still have format: "openai"; fetchOpenAI() rejects any Bedrock secret with Bedrock does not support OpenAI format, so requests to any of these newly advertised Bedrock endpoints fail before reaching AWS. Existing Bedrock-only catalog entries use the converse format, or these should not be exposed as Bedrock endpoints until supported.

Useful? React with 👍 / 👎.

github-actions · 2026-06-19T12:33:24Z

Codex Review Response

PR: #834
Review finding: P1 — Bedrock-only models with incompatible "format": "openai"

Action taken: Changed format from "openai" to "converse" for three Bedrock-only models in packages/proxy/schema/model_list.json:

openai.gpt-5.5 (line 12508)
openai.gpt-5.4 (line 12522)
xai.grok-4.3 (line 13161)

Why: These models have available_providers: ["bedrock"] and are routed exclusively through Bedrock via the MODEL_ENDPOINT_TYPES override in packages/proxy/schema/index.ts. However, fetchOpenAI() in packages/proxy/src/proxy.ts:2131 throws Bedrock does not support OpenAI format when given a Bedrock secret. All other Bedrock-only models in the catalog (e.g., openai.gpt-oss-safeguard-120b, mistral.mistral-large-3-675b-instruct, moonshot.kimi-k2-thinking) correctly use "format": "converse".

No changes were made to packages/proxy/schema/index.ts.

github-actions · 2026-06-20T11:46:10Z

Codex (@codex) review

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: a804561cc8

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".

chatgpt-codex-connector · 2026-06-20T11:50:08Z

+  "openai.gpt-5.5": ["bedrock"],
+  "openai.gpt-5.4": ["bedrock"],
  "nvidia/nemotron-3-ultra-550b-a55b": ["together"],
  "moonshot.kimi-k2-thinking": ["bedrock"],
+  "xai.grok-4.3": ["bedrock"],


Don't advertise Bedrock for OpenAI-format specs

These new endpoint entries make getModelEndpointTypes offer Bedrock for openai.gpt-5.5, openai.gpt-5.4, and xai.grok-4.3, but their catalog specs still have format: "openai" in model_list.json. In this proxy, fetchModel dispatches by modelSpec.format, and fetchOpenAI immediately rejects Bedrock secrets with Bedrock does not support OpenAI format, so selecting these newly advertised Bedrock endpoints fails before reaching AWS. Either keep them out of AvailableEndpointTypes or change the Bedrock specs to the Converse format used by the existing Bedrock OpenAI models.

Useful? React with 👍 / 👎.

chatgpt-codex-connector · 2026-06-20T11:50:08Z

    "reasoning": true,
    "max_input_tokens": 202800,
-    "max_output_tokens": 128000,
+    "max_output_tokens": 202800,


Keep GLM 5.1 output capped below its context window

This copies the 202.8k context window into max_output_tokens; however GLM-5.1's documented maximum max_tokens is 131072 (see Z.AI's core-parameters table), and the sibling zai-org/GLM-5.1 entry above remains capped at 128000. With this value, clients can accept or default to max_tokens values Fireworks will reject for accounts/fireworks/models/glm-5p1; keep 202800 as the input/context limit and leave the output cap around 128K.

Useful? React with 👍 / 👎.

github-actions · 2026-06-20T11:52:45Z

Codex Review Response

Changes made

P1: Bedrock models using wrong format (openai.gpt-5.5, openai.gpt-5.4, xai.grok-4.3)

Changed format from "openai" to "converse" for three Bedrock-only model entries in model_list.json:

openai.gpt-5.5 (line 12508)
openai.gpt-5.4 (line 12522)
xai.grok-4.3 (line 13161)

The proxy's fetchOpenAI rejects Bedrock secrets with "Bedrock does not support OpenAI format" (proxy.ts:2131), so these models would fail at runtime. All other Bedrock models in the catalog use format: "converse", which is the correct dispatch path.

No changes to index.ts were needed — the AvailableEndpointTypes entries correctly advertise Bedrock for these model IDs; the issue was only the format mismatch in the model specs.

P2: GLM 5.1 max_output_tokens on Fireworks

Changed max_output_tokens from 202800 to 128000 for accounts/fireworks/models/glm-5p1 in model_list.json (line 10439). The previous value duplicated the context window size. The sibling entry zai-org/GLM-5.1 already uses 128000, which aligns with the documented maximum.

github-actions · 2026-06-21T12:03:44Z

Codex (@codex) review

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 784981375e

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".

chatgpt-codex-connector · 2026-06-21T12:07:12Z

+  "openai.gpt-5.5": ["bedrock"],
+  "openai.gpt-5.4": ["bedrock"],
  "nvidia/nemotron-3-ultra-550b-a55b": ["together"],
  "moonshot.kimi-k2-thinking": ["bedrock"],
+  "xai.grok-4.3": ["bedrock"],


Use a Bedrock-compatible format for these models

When any of these newly exposed Bedrock mappings is selected, the corresponding model specs still declare format: "openai", so fetchModel dispatches into fetchOpenAI() and that path immediately rejects secret.type === "bedrock" with “Bedrock does not support OpenAI format.” In environments where the credentials lookup returns a Bedrock secret (or fallback secret selection uses getModelEndpointTypes), these advertised Bedrock models will fail before reaching Bedrock; either mark them with a Bedrock-supported format/routing path or do not expose them as Bedrock endpoint types yet.

Useful? React with 👍 / 👎.

github-actions · 2026-06-21T12:11:18Z

Codex Review Response

Issue: Bedrock models with incompatible `format: "openai"`

Valid issue, fixed.

The Codex review correctly identified that some Bedrock-only models had "format": "openai", which causes fetchOpenAI() to throw ProxyBadRequestError("Bedrock does not support OpenAI format") at packages/proxy/src/proxy.ts.

Models fixed

Changed "format": "openai" to "format": "converse" in packages/proxy/schema/model_list.json for:

openai.gpt-5.5 (line 12524) - Bedrock-only, was "openai", now "converse"
openai.gpt-5.4 (line 12538) - Bedrock-only, was "openai", now "converse"
xai.grok-4.3 (line 13177) - Bedrock-only, was "openai", now "converse"

All other Bedrock models already use the correct format ("converse" for non-Anthropic models, "anthropic" for Claude models). No changes were needed in packages/proxy/schema/index.ts.

github-actions · 2026-06-22T13:06:55Z

Codex (@codex) review

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 6b404c619f

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".

chatgpt-codex-connector · 2026-06-22T13:11:08Z

+  "openai.gpt-5.5": ["bedrock"],
+  "openai.gpt-5.4": ["bedrock"],
  "nvidia/nemotron-3-ultra-550b-a55b": ["together"],
  "moonshot.kimi-k2-thinking": ["bedrock"],
+  "xai.grok-4.3": ["bedrock"],


Use a Bedrock-compatible format for these models

In this repo, these newly exposed Bedrock-only IDs are defined in model_list.json with format: "openai", and fetchModel dispatches by that format before considering the provider. When getModelEndpointTypes returns bedrock for these models, a Bedrock secret will be sent to fetchOpenAI, which immediately throws Bedrock does not support OpenAI format in packages/proxy/src/proxy.ts:2130-2131, so requests for these advertised Bedrock models fail before reaching Bedrock; either do not expose them as Bedrock here or give them a Bedrock-supported format/path.

Useful? React with 👍 / 👎.

github-actions · 2026-06-22T13:14:42Z

Codex Review Response

Issue: Bedrock-only models with `format: "openai"` will fail at runtime

Status: Fixed

Codex correctly identified that three Bedrock-only models had format: "openai", which causes fetchOpenAI to throw "Bedrock does not support OpenAI format" before the request reaches Bedrock.

Models fixed in `packages/proxy/schema/model_list.json`

Changed format from "openai" to "converse" for:

openai.gpt-5.5 (line 12523)
openai.gpt-5.4 (line 12537)
xai.grok-4.3 (line 13176)

This is consistent with all other Bedrock-only models from non-Anthropic providers (e.g., openai.gpt-oss-safeguard-120b, moonshot.kimi-k2-thinking, mistral.*), which correctly use format: "converse".

No changes were needed in packages/proxy/schema/index.ts — it only contains provider endpoint mappings, not format data.

…rmat to converse Both model-sync workflows have a 'Respond to Codex review with Claude Code' step that auto-applies Codex's suggestions to the catalog. When the index.ts Bedrock mappings for openai.gpt-5.5 / openai.gpt-5.4 / xai.grok-4.3 are present, Codex posts a P1 ('openai format on a Bedrock model fails') because the TS proxy's fetchOpenAI rejects Bedrock secrets, and the auto-apply step 'fixes' it by flipping format openai->converse. That is wrong: these are Bedrock Mantle models served only via the OpenAI-compatible bedrock-mantle endpoint, which does not support the Converse API or InvokeModel, so converse breaks invocation at AWS. This recurred across #834/#836/#840/#843 (all reverting the validated #822 openai fix). The LiteLLM field sync never touches format (no checkAndUpdateFormat), so this guardrail on the codex-response prompt is the correct lever. Adds a 'Bedrock Mantle exception' instruction to the codex-response prompt in both sync-models.yaml and fix-missing-model-bot-issues.yaml: keep format=openai for bedrock-only openai-format models and explain in the summary that this is an intentional Mantle case to be handled proxy-side, not a catalog change. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

github-actions Bot force-pushed the chore/sync-new-models branch from fefb293 to 40df2b1 Compare June 19, 2026 12:27

github-actions Bot added the auto-sync label Jun 19, 2026

vercel Bot deployed to Preview June 19, 2026 12:28 View deployment

chatgpt-codex-connector Bot reviewed Jun 19, 2026

View reviewed changes

github-actions Bot force-pushed the chore/sync-new-models branch from 40df2b1 to f6ade0e Compare June 19, 2026 12:33

vercel Bot deployed to Preview June 19, 2026 12:34 View deployment

github-actions Bot force-pushed the chore/sync-new-models branch from f6ade0e to a804561 Compare June 20, 2026 11:46

vercel Bot deployed to Preview June 20, 2026 11:46 View deployment

chatgpt-codex-connector Bot reviewed Jun 20, 2026

View reviewed changes

github-actions Bot force-pushed the chore/sync-new-models branch from a804561 to f07127c Compare June 20, 2026 11:52

vercel Bot deployed to Preview June 20, 2026 11:52 View deployment

github-actions Bot force-pushed the chore/sync-new-models branch from f07127c to 7849813 Compare June 21, 2026 12:03

vercel Bot deployed to Preview June 21, 2026 12:04 View deployment

chatgpt-codex-connector Bot reviewed Jun 21, 2026

View reviewed changes

github-actions Bot force-pushed the chore/sync-new-models branch from 7849813 to 6094b11 Compare June 21, 2026 12:11

vercel Bot deployed to Preview June 21, 2026 12:11 View deployment

github-actions Bot force-pushed the chore/sync-new-models branch from 6094b11 to 6b404c6 Compare June 22, 2026 13:06

vercel Bot deployed to Preview June 22, 2026 13:07 View deployment

chatgpt-codex-connector Bot reviewed Jun 22, 2026

View reviewed changes

chore: respond to codex review

3ab5c4c

github-actions Bot force-pushed the chore/sync-new-models branch from 6b404c6 to 3ab5c4c Compare June 22, 2026 13:14

vercel Bot deployed to Preview June 22, 2026 13:14 View deployment

Erin McNulty (erin2722) closed this Jun 22, 2026

Uh oh!

Conversation

github-actions Bot commented Jun 19, 2026

Uh oh!

vercel Bot commented Jun 19, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions Bot commented Jun 19, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector Bot Jun 19, 2026

Choose a reason for hiding this comment

Uh oh!

github-actions Bot commented Jun 19, 2026

Codex Review Response

Uh oh!

github-actions Bot commented Jun 20, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector Bot Jun 20, 2026

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector Bot Jun 20, 2026

Choose a reason for hiding this comment

Uh oh!

github-actions Bot commented Jun 20, 2026

Codex Review Response

Changes made

P1: Bedrock models using wrong format (openai.gpt-5.5, openai.gpt-5.4, xai.grok-4.3)

P2: GLM 5.1 max_output_tokens on Fireworks

Uh oh!

github-actions Bot commented Jun 21, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector Bot Jun 21, 2026

Choose a reason for hiding this comment

Uh oh!

github-actions Bot commented Jun 21, 2026

Codex Review Response

Issue: Bedrock models with incompatible format: "openai"

Models fixed

Uh oh!

github-actions Bot commented Jun 22, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector Bot Jun 22, 2026

Choose a reason for hiding this comment

Uh oh!

github-actions Bot commented Jun 22, 2026

Codex Review Response

Issue: Bedrock-only models with format: "openai" will fail at runtime

Models fixed in packages/proxy/schema/model_list.json

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

vercel Bot commented Jun 19, 2026 •

edited

Loading

Issue: Bedrock models with incompatible `format: "openai"`

Issue: Bedrock-only models with `format: "openai"` will fail at runtime

Models fixed in `packages/proxy/schema/model_list.json`