Providers

OpenAlloy supports 13+ LLM providers through a unified interface. Configure your API keys and switch between providers seamlessly.

Supported Providers

Provider Models Config Key
Anthropic Claude 4.5, Claude 4, Claude 3.5 anthropic.api_key
OpenAI GPT-4o, GPT-4, o1, o3 openai.api_key
Google Gemini 2.5, Gemini 2.0 google.api_key
Mistral Mistral Large, Codestral mistral.api_key
Groq LLaMA 3, Mixtral groq.api_key
Cohere Command R+ cohere.api_key
DeepSeek DeepSeek V3, DeepSeek R1 deepseek.api_key
xAI Grok 3 xai.api_key
Perplexity Sonar perplexity.api_key
Together Open-source models together.api_key
Fireworks Open-source models fireworks.api_key
Ollama Local models ollama.base_url
OpenRouter Multi-provider proxy openrouter.api_key

Switching Providers

In TUI Mode

Press Tab to open the model picker, or type:

/model anthropic:claude-sonnet-4-20250514

In Agent Mode

openalloy agent --provider openai --model gpt-4o -f input.txt

In Config

openalloy config set default.provider anthropic
openalloy config set default.model claude-sonnet-4-20250514

Ollama (Local Models)

OpenAlloy supports locally-hosted models via Ollama:

openalloy config set ollama.base_url http://localhost:11434
openalloy chat --provider ollama --model llama3

Provider-Specific Features

Some providers expose features beyond the unified interface:

  • Anthropic: Extended thinking, tool use, prompt caching
  • OpenAI: Function calling, vision, DALL-E
  • Google: Grounding, code execution
  • DeepSeek: R1 reasoning chains

These are available through provider-specific flags when using openalloy agent.