feat: add new Chutes provider models

[roomote]

This PR adds two new models to the Chutes provider as requested in issue #6698:

• Qwen/Qwen3-Coder-480B-A35B-Instruct-FP8 - A Qwen3 Coder model optimized for coding tasks
• moonshotai/Kimi-K2-Instruct - Moonshot AI Kimi K2 Instruct model

Changes Made:

• Added new model IDs to the ChutesModelId type in packages/types/src/providers/chutes.ts
• Added model configurations with appropriate context windows and descriptions to the chutesModels object
• Added comprehensive tests for both new models in src/api/providers/__tests__/chutes.spec.ts

Testing:

• All existing tests pass ✅
• New tests added for both models ✅
• TypeScript compilation successful ✅
• Linting checks passed ✅

Fixes #6698

---

Important

Add new models Qwen/Qwen3-Coder-480B-A35B-Instruct-FP8 and moonshotai/Kimi-K2-Instruct-75k to Chutes provider with configurations and tests.

• Models:
  • Add Qwen/Qwen3-Coder-480B-A35B-Instruct-FP8 and moonshotai/Kimi-K2-Instruct-75k to ChutesModelId in chutes.ts.
  • Update chutesModels with configurations for new models, including context windows and descriptions.
• Tests:
  • Add tests for new models in chutes.spec.ts to verify correct configurations and behavior.
• Misc:
  • Ensure all existing tests pass and new tests are comprehensive.

^{This description was created by for ec5d8ac. You can customize this summary. It will automatically update as commits are pushed.}

[roomote]

I reviewed my own code and found it only 40% complete. Classic me.

[roomote]

I notice that the issue #6698 requested 5 models, but I only added 2. Looking at the existing code, I see that and are already present (lines 29-30). The issue had duplicates for . Could we update the PR description to clarify that 3 of the 5 requested models were already implemented?

[daniel-lxs]

Not really sure about the pricing or context window since chutes doesn't seem to provide that in the website nor the JSON available in the API

[jordanmiguel]

@daniel-lxs @hannesrudolph can we expedite this? Qwen3 coder is a "mainstream" model for everyone these days. There's even a solution to dynamically fetch the models from chutes in here: #6698 (comment)

[daniel-lxs]

@jordanmiguel I think we can initially add the new models, we just need to get the info of the pricing (which seems to be already available according to the comment you shared but also it would be great to the the context window, do you know where we can find that in the website or API?

[jordanmiguel]

@daniel-lxs yes, if you access the model inside chutes website you can see it on the top right:

0,19999 USD Input
0,8001 USD Output
(Same price for both models)

And it's free with a Chutes subscription (up until you reach the requests limit)
So it's safe to assume 0.00 and 0.00 for Chutes direct provider.

[jordanmiguel]

@daniel-lxs Kimi K2:
Input: 0.1481
Output: 0.5926

[jordanmiguel]

@daniel-lxs About the context window, they don't have that information on the website, but I think it's safe to assume initially that it's the same on openrouter:

Kimi K2:
75k

Qwen3 Coder:
262k

[jordanmiguel]

@daniel-lxs lastly, there's one mistake on this PR.
Model for Kimi-K2 on chutes API is actually "moonshotai/Kimi-K2-Instruct-75k", not "moonshotai/Kimi-K2-Instruct"

[daniel-lxs]

@roomote-agent Adjust the pricing and context window of the models added based on the comments, do not create a new PR

[roomote]

Hi @daniel-lxs! I'm on it. I'll adjust the pricing and context windows for both models based on the information provided in the comments:

• Qwen3 Coder: Context window 262k, pricing $0.00/$0.00 (free with Chutes subscription)
• Kimi K2: Context window 75k, pricing $0.1481/$0.5926, and fix model ID to include "-75k" suffix

Working on these changes now.

[roomote]

✅ @daniel-lxs All requested changes have been completed and pushed to the PR branch!

Changes Made:

• Qwen3 Coder (Qwen/Qwen3-Coder-480B-A35B-Instruct-FP8):

  • ✅ Updated context window from 131k to 262k
  • ✅ Pricing remains at $0.00/$0.00 (free with Chutes subscription)

• Kimi K2 (moonshotai/Kimi-K2-Instruct-75k):

  • ✅ Fixed model ID to include -75k suffix
  • ✅ Updated context window from 131k to 75k
  • ✅ Updated pricing to $0.1481 (input) / $0.5926 (output)

Testing:

• ✅ All unit tests pass
• ✅ TypeScript compilation successful
• ✅ Linting checks pass
• ✅ All CI checks have passed

The models are now configured with the correct specifications based on the information provided in the comments above.

[daniel-lxs]

LGTM