Skip to content

Add new kimi model to groq #7692

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
wants to merge 1 commit into from
Closed

Add new kimi model to groq #7692

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
13 changes: 12 additions & 1 deletion packages/types/src/providers/groq.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,10 +11,11 @@ export type GroqModelId =
| "qwen/qwen3-32b"
| "deepseek-r1-distill-llama-70b"
| "moonshotai/kimi-k2-instruct"
| "moonshotai/kimi-k2-instruct-0905"
| "openai/gpt-oss-120b"
| "openai/gpt-oss-20b"

export const groqDefaultModelId: GroqModelId = "llama-3.3-70b-versatile" // Defaulting to Llama3 70B Versatile
export const groqDefaultModelId: GroqModelId = "moonshotai/kimi-k2-instruct-0905"
Copy link
Contributor

@roomote roomote bot Sep 5, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

⚠️ Breaking Change: Changing the default model from llama-3.3-70b-versatile to moonshotai/kimi-k2-instruct-0905 will affect all users who rely on the default model. They may experience different behavior, costs, and capabilities. Could we consider:

  1. Keeping the existing default for backward compatibility
  2. Or documenting this breaking change prominently in the changelog/release notes?


export const groqModels = {
// Models based on API response: https://api.groq.com/openai/v1/models
Expand Down Expand Up @@ -100,6 +101,16 @@ export const groqModels = {
cacheReadsPrice: 0.5, // 50% discount for cached input tokens
description: "Moonshot AI Kimi K2 Instruct 1T model, 128K context.",
},
"moonshotai/kimi-k2-instruct-0905": {
maxTokens: 16384,
contextWindow: 262144,
supportsImages: false,
supportsPromptCache: true,
inputPrice: 0.6,
Copy link
Contributor

@roomote roomote bot Sep 5, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is the significant price reduction intentional? The new model has:

  • Input price: /bin/sh.6 (40% lower than the existing kimi-k2-instruct at .0)
  • Output price: .5 (17% lower than the existing kimi-k2-instruct at .0)

Could you confirm these prices are correct according to Groq's pricing?

outputPrice: 2.5,
cacheReadsPrice: 0.15,
Copy link
Contributor

@roomote roomote bot Sep 5, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The cache reads price of /bin/sh.15 represents a 75% discount from the input price (/bin/sh.6), while the existing kimi-k2-instruct model has a 50% discount. Is this aggressive discount structure correct?

description: "Moonshot AI Kimi K2 Instruct 1T model, 256K context.",
Copy link
Contributor

@roomote roomote bot Sep 5, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Minor inconsistency: The description says "256K context" but the contextWindow value is 262144 (which is 256Ki or 256 * 1024). While technically correct, this differs from the pattern used in the existing model which says "128K context" with value 131072. Should we use "256Ki context" or keep it consistent with the existing pattern?

},
"openai/gpt-oss-120b": {
maxTokens: 32766,
contextWindow: 131072,
Expand Down
Loading