LLM Providers

Hybrix supports multiple LLM providers out of the box. Each agent can use a different provider, letting you mix models based on cost, speed, and capability.

Anthropic

Access Claude models via the Anthropic API. Set your API key in the environment and configure the agent:

json

{
  "provider": "anthropic",
  "model": "claude-sonnet-4-20250514"
}

Model	Best For	Context
claude-opus-4-20250514	Complex reasoning, research	200K
claude-sonnet-4-20250514	Balanced performance	200K
claude-haiku-3-5-20241022	Fast, low-cost tasks	200K

OpenAI

Use GPT and o-series models via the OpenAI API:

json

{
  "provider": "openai",
  "model": "gpt-4o"
}

Model	Best For	Context
gpt-4o	General purpose, multimodal	128K
gpt-4o-mini	Fast, cost-effective	128K
o3	Advanced reasoning	200K

Google (Gemini)

Access Gemini models via the Google AI Studio API:

json

{
  "provider": "google",
  "model": "gemini-2.5-pro"
}

Local models (Ollama)

Run models locally using Ollama. No API key required. Make sure Ollama is installed and running.

bash

# Pull a model first
ollama pull llama3.1:8b

json

{
  "provider": "local",
  "model": "llama3.1:8b",
  "baseUrl": "http://localhost:11434"
}

Local models run entirely on your hardware. Performance depends on your GPU/CPU. The baseUrl field is optional if Ollama runs on the default port.

OpenAI-compatible endpoints

Any API that follows the OpenAI chat completions format can be used with the custom provider. This works with vLLM, Together AI, Groq, and other providers.

json

{
  "provider": "custom",
  "model": "meta-llama/Llama-3-70b-chat-hf",
  "baseUrl": "https://api.together.xyz/v1",
  "apiKey": "$TOGETHER_API_KEY"
}

← Agent Roles Hybrid Mascots →