fix: add Together models Qwen/Qwen3.5-397B-A17B +8 more

[github-actions]

fix: add Together models Qwen/Qwen3.5-397B-A17B +8 more

Closes #476

Source issue: #476

Summary

Field	Value
Provider	together
Primary model	Qwen/Qwen3.5-397B-A17B
Changed models	Qwen/Qwen3.5-397B-A17B Qwen/Qwen3.5-9B Qwen/Qwen3-Coder-Next-FP8 Qwen/Qwen3-Coder-480B-A35B-Instruct-FP8 MiniMaxAI/MiniMax-M2.7 google/gemma-3n-E4B-it LiquidAI/LFM2-24B-A2B deepcogito/cogito-v2-1-671b essentialai/rnj-1-instruct
Added models	Qwen/Qwen3.5-397B-A17B Qwen/Qwen3.5-9B Qwen/Qwen3-Coder-Next-FP8 Qwen/Qwen3-Coder-480B-A35B-Instruct-FP8 MiniMaxAI/MiniMax-M2.7 google/gemma-3n-E4B-it LiquidAI/LFM2-24B-A2B deepcogito/cogito-v2-1-671b essentialai/rnj-1-instruct
Updated models	None
Verification sources	1 2 3

Verified metadata

Model	Display name	Parent	Providers	Format	Flavor	Token limits	Pricing	Lifecycle
Qwen/Qwen3.5-397B-A17B	Qwen 3.5 397B A17B		together	openai	chat	input=262144, output=not provided	in/out=0.6/3.6 per 1M	multimodal=true
Qwen/Qwen3.5-9B	Qwen 3.5 9B		together	openai	chat	input=262144, output=not provided	in/out=0.1/0.15 per 1M	multimodal=true
Qwen/Qwen3-Coder-Next-FP8	Qwen3 Coder Next FP8		together	openai	chat	input=262144, output=not provided	in/out=0.5/1.2 per 1M	active
Qwen/Qwen3-Coder-480B-A35B-Instruct-FP8	Qwen3 Coder 480B A35B Instruct FP8		together	openai	chat	input=256000, output=not provided	in/out=2/2 per 1M	active
MiniMaxAI/MiniMax-M2.7	MiniMax M2.7		together	openai	chat	input=202752, output=not provided	in/out=0.3/1.2 per 1M; cache read=0.06 per 1M	active
google/gemma-3n-E4B-it	Gemma 3n E4B IT		together	openai	chat	input=32768, output=not provided	in/out=0.06/0.12 per 1M	active
LiquidAI/LFM2-24B-A2B	LFM2 24B A2B		together	openai	chat	input=32768, output=not provided	in/out=0.03/0.12 per 1M	active
deepcogito/cogito-v2-1-671b	Cogito v2.1 671B		together	openai	chat	input=163840, output=not provided	in/out=1.25/1.25 per 1M	active
essentialai/rnj-1-instruct	RNJ-1 Instruct		together	openai	chat	input=32768, output=not provided	in/out=0.15/0.15 per 1M	active

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	Why sync_models may be wrong	Official verification
Qwen/Qwen3.5-9B	catalog entry	present	missing	None	No translated sync_models reference entry matched this model name.	1 2 3
Qwen/Qwen3-Coder-Next-FP8	catalog entry	present	missing	None	No translated sync_models reference entry matched this model name.	1 2 3
MiniMaxAI/MiniMax-M2.7	catalog entry	present	missing	None	No translated sync_models reference entry matched this model name.	1 2 3
google/gemma-3n-E4B-it	catalog entry	present	missing	None	No translated sync_models reference entry matched this model name.	1 2 3
LiquidAI/LFM2-24B-A2B	catalog entry	present	missing	None	No translated sync_models reference entry matched this model name.	1 2 3
deepcogito/cogito-v2-1-671b	catalog entry	present	missing	None	No translated sync_models reference entry matched this model name.	1 2 3
essentialai/rnj-1-instruct	catalog entry	present	missing	None	No translated sync_models reference entry matched this model name.	1 2 3

[vercel]

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

Project	Deployment	Actions	Updated (UTC)
ai-proxy	Ready	Preview, Comment	Apr 28, 2026 5:55pm

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 6f7682b2ac

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]

Add endpoint mapping for newly added Baseten Gemma models

These new entries declare available_providers: ["baseten"], but no corresponding keys were added to AvailableEndpointTypes in packages/proxy/schema/index.ts; getModelEndpointTypes() therefore falls back to the OpenAI default (["openai", "azure"]) for these model names. In the edge fallback path (edge/index.ts), that makes the proxy choose an OpenAI/Azure secret for Gemma 4 Baseten models when key lookup fails, causing requests to be routed with the wrong provider credentials.

Useful? React with 👍 / 👎.