🦞OpenClawThe board builders scan before they commit to this stack.

Check your current version

If you see 3.10.x or 3.11.x, skip to step 2. If you see 3.8 or 3.9, or "command not found", keep going.

Install Python 3.11

Mac: brew install [email protected] (install Homebrew first if you don't have it)

Windows: Download from python.org. Check "Add Python to PATH" during install.

Linux: sudo apt install python3.11 python3.11-venv

Create the environment

This creates a folder called interpreter_env with its own Python. Your system Python stays untouched.

Activate it

Mac/Linux:

Windows:

You'll see (interpreter_env) in your terminal prompt. That means it's working.

Install via pip

Make sure your virtual environment is activated first (you should see (interpreter_env) in your prompt).

Verify it installed

You should see a version number. If you get "command not found", try python3 -m interpreter instead.

Get an OpenAI API key

Go to platform.openai.com/api-keys and create one. You'll need to add a payment method — but costs are low (~$0.01-0.05 per conversation).

Set the key in your terminal

Mac/Linux (add to ~/.zshrc or ~/.bashrc to persist):

Windows:

Save money with GPT-3.5

GPT-4 is the default and costs ~10x more. GPT-3.5 is fine for most tasks. You can also use interpreter --fast as a shortcut.

Launch it

You'll see a chat prompt. Type naturally — it understands plain English.

Things to try

It writes and runs code to accomplish each task. You'll be asked to approve before it executes anything.

Auto-approve mode (skip confirmations)

The -y flag auto-approves all code execution. Only use this when you trust the task — it can modify files and run shell commands.

Install Ollama

Download from ollama.com. One-click install on Mac/Windows/Linux.

Pull a model

Llama 3.1 8B is a great starting point (~4.7GB download). For better results with more RAM, try ollama pull llama3.1:70b.

Run Open Interpreter with your local model

It auto-detects Ollama and uses your downloaded model. Zero API costs. Fully private — nothing leaves your machine.