Native macOS installation via Homebrew and Node.js — free, private, with full Apple Silicon support.
Install OpenClaw locally on Mac — completely free with full data privacy. Supports Apple Silicon (M1/M2/M3/M4) and Intel Macs. Set up in 15 minutes with Homebrew and Node.js. A Mac Mini is a popular 'sweet spot' setup for many users.
Install Node.js 22 LTS using Homebrew:
brew install node@22
Verify the installation:
node --version\nnpm --version
Clone the OpenClaw source code from GitHub:
git clone https://github.com/openclaw/openclaw.git
Navigate to the project and install all dependencies (may take a few minutes):
cd openclaw && npm install
Copy the example environment file and edit it with your API keys:
cp .env.example .env
Open .env with your preferred editor (VS Code, nano, vim, etc.) and set at least one model API key:
# OpenClaw 环境变量配置 OPENAI_API_KEY=sk-your-openai-api-key # ANTHROPIC_API_KEY=sk-ant-your-key # DEEPSEEK_API_KEY=your-deepseek-key DEFAULT_MODEL=gpt-4o ACCESS_CODE=your-secure-password PORT=3210
Start the OpenClaw service:
npm start
Once started, open your browser and visit:
http://localhost:3210
OpenClaw fully supports Apple Silicon (M1, M2, M3, M4 series). The installation process is identical to Intel Macs. All npm dependencies provide native ARM64 builds — no Rosetta translation needed.
OpenClaw runs excellently on Apple Silicon Macs. Thanks to the unified memory architecture and efficient ARM instruction set, local inference is faster and more power-efficient. A Mac Mini (M4) + OpenClaw is widely regarded as the 'sweet spot' setup.
If you plan to use your Mac as a long-term server, keep it plugged in and disable 'Automatic Sleep' in System Settings to prevent service interruption.
Your agent is installed but can't think or talk yet — configure an LLM API and messaging tool to bring it alive.
Feed Your Agent