fix(models): update Cerebras and Groq model definitions

[waleedlatif1]

Summary

• Updated Cerebras models based on current API docs (removed deprecated models, added gpt-oss-120b, qwen-3-32b, qwen-3-235b, zai-glm-4.7)
• Updated Groq models with correct pricing and added missing models (qwen3-32b, gpt-oss-safeguard-20b, deepseek-r1-distill-qwen-32b)
• Fixed model IDs and context windows based on official documentation

Type of Change

• Bug fix

Testing

Tested manually

Checklist

• Code follows project style guidelines
• Self-reviewed my changes
• Tests added/updated and passing
• No new warnings introduced
• I confirm that I have read and agree to the terms outlined in the Contributor License Agreement (CLA)

[vercel]

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

1 Skipped Deployment

Project	Deployment	Review	Updated (UTC)
docs	Skipped		Jan 28, 2026 2:03am

[greptile-apps]

Greptile Overview

Greptile Summary

Updated model definitions for Cerebras and Groq providers to align with current API documentation. Changes include:

Cerebras:

• Changed default model from llama-3.3-70b to gpt-oss-120b
• Fixed model ID format: llama-3.1-8b → llama3.1-8b (removed dash)
• Removed deprecated models: llama-3.1-70b, llama-4-scout-17b-16e-instruct
• Added new models: gpt-oss-120b, qwen-3-32b, qwen-3-235b-a22b-instruct-2507, zai-glm-4.7
• Updated pricing for llama-3.3-70b (input: $0.60 → $0.85, output: $0.60 → $1.20)
• Updated context windows to 131000 for new models

Groq:

• Corrected pricing for gpt-oss-120b (output: $0.75 → $0.60)
• Corrected pricing for gpt-oss-20b (input: $0.01 → $0.075, output: $0.25 → $0.30)
• Added new models: gpt-oss-safeguard-20b, qwen3-32b, deepseek-r1-distill-qwen-32b
• Fixed model ID: llama-4-scout-17b-instruct → meta-llama/llama-4-scout-17b-16e-instruct
• Removed duplicate: llama-4-maverick-17b-instruct (kept the 128e variant)
• Updated pricing for llama-4-maverick-17b-128e-instruct (input: $0.50 → $0.20, output: $0.77 → $0.60)
• Fixed model ID and context window for kimi-k2-instruct → kimi-k2-instruct-0905 (contextWindow: 131072 → 262144)

All updatedAt timestamps refreshed to 2026-01-27.

Confidence Score: 5/5

• This PR is safe to merge - it only updates model definitions with current API documentation
• The changes are straightforward configuration updates to model definitions, pricing, and IDs based on official API documentation. No logic changes, no security concerns, and the modifications align with the PR description. All changes are isolated to the models configuration file.
• No files require special attention

Important Files Changed

Filename	Overview
apps/sim/providers/models.ts	Updated Cerebras and Groq model definitions with current pricing, added new models, removed deprecated models, and fixed model IDs based on official API documentation

Sequence Diagram

sequenceDiagram
    participant User
    participant App
    participant Models as models.ts
    participant API as LLM Provider API
    
    Note over Models: Model Definition Updates
    
    User->>App: Select LLM model
    App->>Models: Get available models
    
    alt Cerebras Provider
        Models-->>App: Return updated models<br/>(gpt-oss-120b as default,<br/>qwen-3-32b, qwen-3-235b, zai-glm-4.7)
        App->>API: Call with updated model ID<br/>(e.g., cerebras/llama3.1-8b)
        API-->>App: Process request with correct pricing
    else Groq Provider
        Models-->>App: Return updated models<br/>(gpt-oss-safeguard-20b, qwen3-32b,<br/>deepseek-r1-distill-qwen-32b)
        App->>API: Call with corrected model ID<br/>(e.g., groq/meta-llama/llama-4-scout-17b-16e-instruct)
        API-->>App: Process request with corrected pricing
    end
    
    Note over Models,API: Pricing calculations now<br/>reflect current API costs

Loading

[greptile-apps]

_{No files reviewed, no comments}

_{Edit Code Review Agent Settings | Greptile}