chore: update model catalog from bot issues

[github-actions]

Automated daily batch of model catalog updates from bot issues.

Included issues

• Closes [BOT ISSUE] Missing Bedrock model IDs for Mistral Large 3, Pixtral Large, and Ministral models #730: [BOT ISSUE] Missing Bedrock model IDs for Mistral Large 3, Pixtral Large, and Ministral models
• Closes [BOT ISSUE] Missing Bedrock model ID for Kimi K2 Thinking (moonshot.kimi-k2-thinking) #731: [BOT ISSUE] Missing Bedrock model ID for Kimi K2 Thinking (moonshot.kimi-k2-thinking)

Summary

Issue	Provider	Primary model	Changed models	Added models	Updated models	Verification sources
#730	bedrock	mistral.mistral-large-3-675b-instruct	mistral.mistral-large-3-675b-instruct mistral.pixtral-large-2502-v1:0 mistral.ministral-3-14b-instruct mistral.ministral-3-8b-instruct mistral.ministral-3-3b-instruct us.mistral.pixtral-large-2502-v1:0 eu.mistral.pixtral-large-2502-v1:0	mistral.mistral-large-3-675b-instruct mistral.pixtral-large-2502-v1:0 mistral.ministral-3-14b-instruct mistral.ministral-3-8b-instruct mistral.ministral-3-3b-instruct us.mistral.pixtral-large-2502-v1:0 eu.mistral.pixtral-large-2502-v1:0	None	1 2 3 4 5 6
#731	bedrock	moonshot.kimi-k2-thinking	moonshot.kimi-k2-thinking moonshotai.kimi-k2-thinking	moonshot.kimi-k2-thinking moonshotai.kimi-k2-thinking	None	1 2

Verified metadata

#730: [BOT ISSUE] Missing Bedrock model IDs for Mistral Large 3, Pixtral Large, and Ministral models

Model	Display name	Parent	Providers	Format	Flavor	Token limits	Pricing	Lifecycle
mistral.mistral-large-3-675b-instruct	Mistral Large 3 (675B)		bedrock	converse	chat	input=256000, output=32000	in/out=0.5/1.5 per 1M	multimodal=true
mistral.pixtral-large-2502-v1:0	Pixtral Large (2502)		bedrock	converse	chat	input=128000, output=16000	n/a	multimodal=true
mistral.ministral-3-14b-instruct	Ministral 3 14B		bedrock	converse	chat	input=128000, output=8192	in/out=0.2/0.2 per 1M	multimodal=true
mistral.ministral-3-8b-instruct	Ministral 3 8B		bedrock	converse	chat	input=128000, output=8192	in/out=0.15/0.15 per 1M	multimodal=true
mistral.ministral-3-3b-instruct	Ministral 3 3B		bedrock	converse	chat	input=128000, output=8192	in/out=0.1/0.1 per 1M	multimodal=true
us.mistral.pixtral-large-2502-v1:0	US Pixtral Large (2502)	mistral.pixtral-large-2502-v1:0	bedrock	converse	chat	input=128000, output=16000	n/a	parent=mistral.pixtral-large-2502-v1:0; multimodal=true
eu.mistral.pixtral-large-2502-v1:0	EU Pixtral Large (2502)	mistral.pixtral-large-2502-v1:0	bedrock	converse	chat	input=128000, output=16000	n/a	parent=mistral.pixtral-large-2502-v1:0; multimodal=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
mistral.mistral-large-3-675b-instruct	max_input_tokens	256000	128000	mistral.mistral-large-3-675b-instruct
mistral.mistral-large-3-675b-instruct	max_output_tokens	32000	8192	mistral.mistral-large-3-675b-instruct
mistral.pixtral-large-2502-v1:0	catalog entry	present	missing	None
us.mistral.pixtral-large-2502-v1:0	max_output_tokens	16000	4096	us.mistral.pixtral-large-2502-v1:0
us.mistral.pixtral-large-2502-v1:0	input_cost_per_mil_tokens	n/a	2	us.mistral.pixtral-large-2502-v1:0
us.mistral.pixtral-large-2502-v1:0	output_cost_per_mil_tokens	n/a	6	us.mistral.pixtral-large-2502-v1:0
eu.mistral.pixtral-large-2502-v1:0	max_output_tokens	16000	4096	eu.mistral.pixtral-large-2502-v1:0
eu.mistral.pixtral-large-2502-v1:0	input_cost_per_mil_tokens	n/a	2	eu.mistral.pixtral-large-2502-v1:0
eu.mistral.pixtral-large-2502-v1:0	output_cost_per_mil_tokens	n/a	6	eu.mistral.pixtral-large-2502-v1:0

#731: [BOT ISSUE] Missing Bedrock model ID for Kimi K2 Thinking (moonshot.kimi-k2-thinking)

Model	Display name	Parent	Providers	Format	Flavor	Token limits	Pricing	Lifecycle
moonshot.kimi-k2-thinking	Kimi K2 Thinking		bedrock	converse	chat	input=256000, output=16384	in/out=0.6/2.5 per 1M	multimodal=false; reasoning=true
moonshotai.kimi-k2-thinking	Kimi K2 Thinking		bedrock	converse	chat	input=256000, output=16384	in/out=0.6/2.5 per 1M	multimodal=false; 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
moonshot.kimi-k2-thinking	max_input_tokens	256000	128000	moonshot.kimi-k2-thinking
moonshot.kimi-k2-thinking	max_output_tokens	16384	8192	moonshot.kimi-k2-thinking
moonshotai.kimi-k2-thinking	catalog entry	present	missing	None

[vercel]

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

Project	Deployment	Actions	Updated (UTC)
ai-proxy	Ready	Preview, Comment	Jun 3, 2026 1:27pm

[github-actions]

Codex (@codex) review

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 4e514ab1ef

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]

Remove the mantle-only Kimi ID from Bedrock runtime catalog

This adds moonshotai.kimi-k2-thinking as a bedrock/converse model, but the proxy's Converse path sends the request through BedrockRuntimeClient with modelId: model; AWS documents moonshot.kimi-k2-thinking as the bedrock-runtime/Converse ID and moonshotai.kimi-k2-thinking as the separate bedrock-mantle Chat Completions ID (https://docs.aws.amazon.com/bedrock/latest/userguide/model-card-moonshot-ai-kimi-k2-thinking.html). Any user selecting this newly exposed alias with a normal Bedrock secret will therefore get a Bedrock runtime validation failure instead of reaching the model.

Useful? React with 👍 / 👎.