chore: update model catalog from bot issues

[github-actions]

Automated daily batch of model catalog updates from bot issues.

Included issues

• Closes [BOT ISSUE] Cerebras: add missing gemma-4-31b model #883: [BOT ISSUE] Cerebras: add missing gemma-4-31b model
• Closes [BOT ISSUE] Anthropic: update Claude Sonnet 4.6 max_output_tokens from 64000 to 128000 #884: [BOT ISSUE] Anthropic: update Claude Sonnet 4.6 max_output_tokens from 64000 to 128000
• Closes [BOT ISSUE] Mistral: add deprecation markers for retired pixtral-large-2411 and mistral-large-2411 #885: [BOT ISSUE] Mistral: add deprecation markers for retired pixtral-large-2411 and mistral-large-2411

Summary

Issue	Provider	Primary model	Changed models	Added models	Updated models	Verification sources
#883	cerebras	gemma-4-31b	gemma-4-31b	gemma-4-31b	None	1 2
#884	anthropic	claude-sonnet-4-6	claude-sonnet-4-6 us.anthropic.claude-sonnet-4-6 global.anthropic.claude-sonnet-4-6 anthropic.claude-sonnet-4-6 publishers/anthropic/models/claude-sonnet-4-6	None	claude-sonnet-4-6 us.anthropic.claude-sonnet-4-6 global.anthropic.claude-sonnet-4-6 anthropic.claude-sonnet-4-6 publishers/anthropic/models/claude-sonnet-4-6	1 2 3
#885	mistral	pixtral-large-2411	pixtral-large-2411 mistral-large-2411	None	pixtral-large-2411 mistral-large-2411	1

Verified metadata

#883: [BOT ISSUE] Cerebras: add missing gemma-4-31b model

Model	Display name	Parent	Providers	Format	Flavor	Token limits	Pricing	Lifecycle
gemma-4-31b	Gemma 4 31B		cerebras	openai	chat	input=131072, output=40960	in/out=0.99/1.49 per 1M	multimodal=true; reasoning=true

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.

Model	Field	Proposed update	sync_models	sync_models source models
gemma-4-31b	catalog entry	present	missing	None

#884: [BOT ISSUE] Anthropic: update Claude Sonnet 4.6 max_output_tokens from 64000 to 128000

Model	Display name	Parent	Providers	Format	Flavor	Token limits	Pricing	Lifecycle
claude-sonnet-4-6	Claude Sonnet 4.6		anthropic	anthropic	chat	input=1000000, output=128000	in/out=3/15 per 1M; cache read=0.3 per 1M; cache write=3.75 per 1M	multimodal=true; reasoning=true
us.anthropic.claude-sonnet-4-6	US Claude 4.6 Sonnet	anthropic.claude-sonnet-4-6	bedrock	anthropic	chat	input=1000000, output=128000	in/out=3.3/16.5 per 1M; cache read=0.33 per 1M; cache write=4.125 per 1M	parent=anthropic.claude-sonnet-4-6; multimodal=true; reasoning=true
global.anthropic.claude-sonnet-4-6	Global Claude 4.6 Sonnet	anthropic.claude-sonnet-4-6	bedrock	anthropic	chat	input=1000000, output=128000	in/out=3/15 per 1M; cache read=0.3 per 1M; cache write=3.75 per 1M	parent=anthropic.claude-sonnet-4-6; multimodal=true; reasoning=true
anthropic.claude-sonnet-4-6	Claude Sonnet 4.6		bedrock	anthropic	chat	input=1000000, output=128000	in/out=3/15 per 1M; cache read=0.3 per 1M; cache write=3.75 per 1M	multimodal=true; reasoning=true
publishers/anthropic/models/claude-sonnet-4-6	Claude Sonnet 4.6		vertex	anthropic	chat	input=1000000, output=128000	in/out=3/15 per 1M; cache read=0.3 per 1M; cache write=3.75 per 1M	multimodal=true; reasoning=true

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.

Model	Field	Proposed update	sync_models	sync_models source models
claude-sonnet-4-6	max_output_tokens	128000	64000	claude-sonnet-4-6
us.anthropic.claude-sonnet-4-6	max_output_tokens	128000	64000	us.anthropic.claude-sonnet-4-6
global.anthropic.claude-sonnet-4-6	max_output_tokens	128000	64000	global.anthropic.claude-sonnet-4-6
anthropic.claude-sonnet-4-6	max_output_tokens	128000	64000	anthropic.claude-sonnet-4-6
publishers/anthropic/models/claude-sonnet-4-6	catalog entry	present	missing	None

#885: [BOT ISSUE] Mistral: add deprecation markers for retired pixtral-large-2411 and mistral-large-2411

Model	Display name	Parent	Providers	Format	Flavor	Token limits	Pricing	Lifecycle
pixtral-large-2411		pixtral-large-latest	mistral	openai	chat	input=128000, output=128000	in/out=2/6 per 1M	parent=pixtral-large-latest; date=2026-05-31; multimodal=true
mistral-large-2411		mistral-large-latest	mistral	openai	chat	input=128000, output=128000	in/out=2/6 per 1M	parent=mistral-large-latest; date=2026-05-31

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.

Model	Field	Proposed update	sync_models	sync_models source models
pixtral-large-2411	deprecation_date	2026-05-31	n/a	mistral/pixtral-large-2411
mistral-large-2411	deprecation_date	2026-05-31	n/a	mistral/mistral-large-2411

[vercel]

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

Project	Deployment	Actions	Updated (UTC)
ai-proxy	Ready	Preview, Comment	Jun 30, 2026 11:19am

[github-actions]

Codex (@codex) review

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 5a1b5ba452

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]

Don't force Vertex Claude 4.6 to global

For Vertex users who set a secret metadata.location such as us-east5, europe-west1, or asia-southeast1 for data residency or Provisioned Throughput, this locations entry makes resolveVertexLocation ignore the configured location (packages/proxy/src/proxy.ts:2646) and always route publishers/anthropic/models/claude-sonnet-4-6 to global. Google's current Claude Sonnet 4.6 model page lists those regional endpoints in addition to the global endpoint (https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/claude/sonnet-4-6), so this should either include all supported locations or omit locations so customer metadata continues to control routing.

Useful? React with 👍 / 👎.

[github-actions]

Codex Review Response

Issue: locations: ["global"] on Vertex Claude Sonnet 4.6

Codex flagged: The locations: ["global"] entry on publishers/anthropic/models/claude-sonnet-4-6 forces resolveVertexLocation (proxy.ts:2646) to always route to the global endpoint, overriding user-configured metadata.location for data residency or Provisioned Throughput.

Action taken: Removed the locations field from publishers/anthropic/models/claude-sonnet-4-6 in model_list.json. This is consistent with all other Vertex Claude models (claude-opus-4-7, claude-opus-4-6, claude-sonnet-4-5, etc.), none of which specify locations. Without the field, resolveVertexLocation falls through to the user's secret metadata location, then to the default (us-central1).

No changes were needed in packages/proxy/schema/index.ts.