Supported LLMs

Site admins can configure Cody via modelConfiguration on a Sourcegraph Enterprise instance. See Model Configuration for more details.

Chat and Prompts

Cody supports a variety of cutting-edge large language models for use in chat and prompts, allowing you to select the best model for your use case.

ProviderModelStatusVision Support
AnthropicClaude Opus 4.7✅✅
AnthropicClaude Opus 4.6✅✅
AnthropicClaude Sonnet 4.6✅✅
AnthropicClaude Sonnet 4.6 with Thinking✅✅
AnthropicClaude Sonnet 4.5✅✅
AnthropicClaude Sonnet 4.5 with Thinking✅✅
AnthropicClaude Opus 4.5✅✅
AnthropicClaude Opus 4.5 with Thinking✅✅
AnthropicClaude Haiku 4.5✅✅
AnthropicClaude Haiku 4.5 with Thinking✅✅
GoogleGemini 2.5 Flash✅✅
GoogleGemini 2.5 Pro✅✅
GoogleGemini 3 Flash✅❌
GoogleGemini 3.1 Flash Lite✅❌
GoogleGemini 3.1 Pro✅ (beta)❌
OpenAIGPT-5.4✅✅
OpenAIGPT-5.4 mini✅✅
OpenAIGPT-5.4 nano✅✅
OpenAIGPT-5.2✅✅
OpenAIGPT-5.1✅✅
OpenAIGPT-5✅✅
OpenAIGPT-5 mini✅✅
OpenAIGPT-5 nano✅✅
OpenAIGPT-4o✅✅
OpenAIGPT-4.1✅✅
OpenAIGPT-4o-mini✅✅
OpenAIGPT-4.1-mini✅✅
OpenAIGPT-4.1-nano✅✅
OpenAIo3✅❌
OpenAIo4-mini✅❌

While Gemini models support vision capabilities, Cody clients do not currently support image uploads to Gemini models.

Autocomplete

Cody uses a set of models for autocomplete which are suited for the low latency use case.

ProviderModelStatusHosting region
AnthropicClaude Haiku 4.5✅global
AnthropicClaude Haiku 4.5 with Thinking✅global
Fireworks.aiDeepSeek V2 Lite Base✅US Iowa
Fireworks.aiAutoedits Fine-Tune DeepSeek Coder V2✅US Iowa
Fireworks.aiNLS Query Translator✅US Iowa
OpenAIGPT-4.1-nano✅global

Smart Apply

ProviderModelStatusHosting region
Fireworks.aiSmart Apply Fine-Tune Qwen 32B V1✅US Iowa

Default Models

The following models are used by default for each feature when no specific model is configured:

FeatureDefault Model
ChatClaude Sonnet 4.5
AutocompleteDeepSeek V2 Lite Base
Fast ChatClaude Haiku 4.5
Previous
Run Cody via proxies
Next
Feature parity reference for Cody clients