Run Claude Code with free local AI (Ollama)

Install Ollama

Open a terminal and run the command for your system. You can also download the installer from ollama.com.

irm https://ollama.com/install.ps1 | iex

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

Download a model and test it

This downloads Google’s Gemma 4 and opens a chat in your terminal. Type a message to check it works, then /bye to exit. Smaller models run on a normal laptop; bigger ones need more memory.

ollama run gemma4

Install Claude Code

Skip this if you already have it. Then check it works with claude --version.

irm https://claude.ai/install.ps1 | iex

curl -fsSL https://claude.ai/install.sh | bash

Connect Claude Code to Ollama

One command. Ollama starts Claude Code and points it at your model through its Anthropic-compatible API.

ollama launch claude

Prefer doing it by hand? Set these, then start Claude Code with your model (example from Ollama’s docs):

Manual setup (optional)

macOS / Linux shell. On Windows PowerShell use $env:ANTHROPIC_AUTH_TOKEN="ollama" and so on.

export ANTHROPIC_AUTH_TOKEN=ollama
export ANTHROPIC_API_KEY=""
export ANTHROPIC_BASE_URL=http://localhost:11434
claude --model qwen3.5

Give the model enough context

For coding tools, Ollama recommends a context window of at least 64,000 tokens so the model can keep your project in mind. In the Ollama app, use the context length slider in Settings, or start the server like this. Bigger context uses more memory; check with ollama ps.

OLLAMA_CONTEXT_LENGTH=64000 ollama serve