Skip to main content

Local Ollama

Run powerful AI models directly on your machine for maximum privacy and zero cost.

Installation​

  1. Download and install Ollama from ollama.com.
  2. Pull a recommended coding model: 
    ollama pull qwen2.5-coder:7b

Auto-Discovery​

The extension automatically detects your local Ollama instance and fetches all installed models. You can see them in the model selection menu.

Custom Endpoint​

If your Ollama instance is running on a different machine or port, you can configure the endpoint in settings:

  • Setting: ollama-code-review.endpoint
  • Default: http://localhost:11434/api/generate
  • Qwen2.5-Coder: Excellent all-around performance for coding tasks.
  • DeepSeek-Coder-V2: Highly capable for complex logic.
  • Llama 3.1 / 3.2: Good for general code reviews and explanations.
  • Codellama: A classic choice specifically tuned for code.