Ollama Integration

Overview

Ollama lets you run AI models locally on your own hardware. No API key needed, full privacy.

Available Models

• Llama 3.1 — Meta's open-source model
• Mistral — Fast and efficient
• Qwen 2.5 — Alibaba's multilingual model
• Deepseek — Strong coding model
• Phi-3 — Microsoft's compact model

Install Ollama

# macOS / Linux
curl -fsSL https://ollama.com/install.sh | sh

# Pull a model
ollama pull llama3.1

Configuration

{
  "agents": {
    "default": {
      "provider": "ollama",
      "model": "llama3.1"
    }
  }
}

Environment Variables

export OLLAMA_BASE_URL="http://localhost:11434"

Next Steps

• Integration Overview — All supported providers
• AI Agent — Agent configuration