How to Install OpenClaw 2026 — Linux, Raspberry Pi, Docker, ClawBox hardware setup guide
🦀 Get ClawBox — OpenClaw, ready out of the box in 5 minutes · 15W · one-time €549 → clawbox.tech
OpenClaw runs on Linux (Ubuntu/Debian), macOS, and comes pre-installed on ClawBox hardware.
The fastest path. ClawBox ships with OpenClaw pre-installed and ready.
- Plug in power and ethernet
- Open the OpenClaw app on your phone
- Scan the QR code on the device
- Follow the setup wizard
Done. No command line needed.
# Ubuntu 22.04 or later
sudo apt update && sudo apt install -y curl git# Download and run the installer
curl -fsSL https://get.openclaw.ai | bash
# Or manual install
git clone https://github.com/openclaw/openclaw.git
cd openclaw
npm install
npm run buildopenclaw start
# Opens setup wizard at http://localhost:3000
# Or run headless:
openclaw setup --headlessopenclaw config set plugins.telegram.token YOUR_BOT_TOKEN
openclaw restartWorks on 64-bit Raspberry Pi OS.
# Install Node.js 20+
curl -fsSL https://deb.nodesource.com/setup_20.x | sudo -E bash -
sudo apt install -y nodejs
# Install OpenClaw
curl -fsSL https://get.openclaw.ai | bashPerformance note: Raspberry Pi 5 runs ~2 tok/s for local inference. Fine for cloud AI (Claude, GPT-4), slow for local models.
# docker-compose.yml
version: '3.8'
services:
openclaw:
image: openclaw/openclaw:latest
ports:
- "3000:3000"
volumes:
- ./workspace:/workspace
- ./openclaw.json:/app/openclaw.json
environment:
- ANTHROPIC_API_KEY=${ANTHROPIC_API_KEY}
- TELEGRAM_BOT_TOKEN=${TELEGRAM_BOT_TOKEN}
restart: unless-stoppeddocker-compose up -dopenclaw.json — main config file:
{
"agents": {
"defaults": {
"model": "anthropic/claude-sonnet-4-6",
"timeoutSeconds": 120
}
},
"plugins": {
"telegram": {
"token": "YOUR_BOT_TOKEN",
"authorizedUsers": [123456789]
}
}
}- Anthropic:
claude-opus-4-6,claude-sonnet-4-6,claude-haiku-4-5 - OpenAI:
openai/gpt-4o,openai/gpt-4o-mini - Google:
google/gemini-2.0-flash - DeepSeek:
deepseek/deepseek-chat - Local (Ollama):
ollama/llama3.1:8b
- Set AI model API key
- Connect at least one messaging channel
- Create
workspace/SOUL.mdwith your assistant's personality - Test with a simple message: "What time is it?"
- Enable heartbeat for proactive check-ins
Port 3000 in use: openclaw config set gateway.port 3001
Telegram bot not responding: Ensure getUpdates is working — openclaw plugins status
Out of memory (Raspberry Pi): Add swap space or use cloud models only
Full docs: docs.openclaw.ai
| Use Case | Recommended | Notes |
|---|---|---|
| Just starting | Raspberry Pi 5 8GB | €80, slow local inference |
| Serious use | ClawBox | €549, 67 TOPS, plug-and-play |
| Power user | Mac Mini M4 16GB | €899, fast local inference |
| Server rack | Ubuntu VPS (4GB+) | Cloud AI only |
→ Fastest setup: ClawBox at clawbox.tech
Read the full guide on ClawBox: OpenClaw setup & install guide →