How to Run OpenClaw on a Raspberry Pi

The Raspberry Pi is the perfect platform for a personal AI agent. Low power, always-on, and completely under your control. If you want to self-host OpenClaw without a power-hungry server, the Pi is your answer.

This guide walks you through installing and optimizing OpenClaw on a Raspberry Pi 4. By the end, you'll have a fully functional AI agent running on $50 hardware.

What You'll Need

Raspberry Pi 4 (4GB or 8GB RAM)
MicroSD card (32GB+) or USB SSD
Power supply (USB-C, 3A+)
Internet connection (WiFi or Ethernet)
OpenAI API key

Note: Raspberry Pi 3 will work but is noticeably slower. Pi Zero and Pi 2 are not recommended due to insufficient RAM.

Why Run on Raspberry Pi?

Low Power

5-7 watts vs 50-100W for a desktop. Perfect for 24/7 operation.

Affordable

$50-100 total cost vs $500+ for a server or ongoing cloud fees.

Private

Your data stays on your hardware. No cloud dependencies.

Always-On

No sleep mode. Your agent is ready whenever you need it.

Step 1: Prepare Your Pi

Start with a fresh Raspberry Pi OS (64-bit) installation:

Download Raspberry Pi Imager
Install Raspberry Pi OS (64-bit) on your SD card/SSD
Boot the Pi and complete initial setup
Enable SSH: sudo raspi-config → Interface Options → SSH
Update system: sudo apt update && sudo apt upgrade -y

Step 2: Install OpenClaw

Install Node.js and OpenClaw:

# Install Node.js 20.x
curl -fsSL https://deb.nodesource.com/setup_20.x | sudo -E bash -
sudo apt install -y nodejs

# Verify installation
node --version  # Should show v20.x.x
npm --version

# Install OpenClaw globally
sudo npm install -g openclaw

# Verify installation
openclaw --version

Step 3: Configure OpenClaw

Set up your agent configuration:

# Create agent directory
mkdir ~/my-agent
cd ~/my-agent

# Initialize OpenClaw
openclaw init

# Set your OpenAI API key
export OPENAI_API_KEY=your-api-key-here

# Edit configuration
nano openclaw.yaml

Add your API key to ~/.bashrc for persistence:

echo 'export OPENAI_API_KEY=your-api-key-here' >> ~/.bashrc
source ~/.bashrc

Step 4: Optimize for Pi

Raspberry Pi has limited resources. Optimize OpenClaw for better performance. See our memory optimization guide for detailed tips.

# openclaw.yaml optimizations for Pi
memory:
  conversation:
    max_messages: 10        # Reduce for Pi's limited RAM
    max_tokens: 2000
    
model:
  default: gpt-3.5-turbo    # Faster, cheaper than GPT-4
  
timeout:
  request: 30000           # 30 second timeout

Step 5: Run Your Agent

Start your agent:

# Run in foreground
openclaw run

# Or run with PM2 for auto-restart
sudo npm install -g pm2
pm2 start openclaw --name my-agent
pm2 save
pm2 startup

Connect to your Pi agent via Telegram or Discord to control it from anywhere.

Troubleshooting

Slow Performance?

Switch to USB SSD if using SD card
Use GPT-3.5 instead of GPT-4
Reduce conversation history limits
Check CPU temperature: vcgencmd measure_temp

Out of Memory?

Increase swap: sudo dphys-swapfile swapoff && sudo nano /etc/dphys-swapfile
Run fewer agents simultaneously
Close unnecessary system services

FAQ

Frequently Asked Questions

Which Raspberry Pi models work?

Raspberry Pi 4 (4GB or 8GB) is recommended. Pi 3 can work but will be slower. Pi Zero and older models lack sufficient RAM and CPU power.

How much power does it use?

OpenClaw on a Pi 4 uses approximately 5-7 watts under normal load. This is significantly more efficient than running on a desktop or cloud VM for 24/7 operation.

Can I run multiple agents?

Yes, but monitor RAM usage. Each agent process uses 200-400MB. On a 4GB Pi, you can comfortably run 2-3 agents. On 8GB, 4-5 agents.

Is it fast enough?

Response times are 2-4 seconds for typical queries. Slower than desktop but acceptable for many use cases. The Pi handles the orchestration; LLM calls still go to APIs.

Should I use SD card or SSD?

SSD strongly recommended. SD cards are slow and wear out quickly with frequent writes. A USB SSD dramatically improves performance and reliability.

Final Thoughts

Running OpenClaw on a Raspberry Pi is a practical way to self-host your AI agent. It's not as fast as a desktop or cloud server, but for $50 you get a private, always-on assistant that uses minimal power.

The Pi shines for personal use, home automation integrations, and learning. For production workloads with many users, consider managed hosting or a more powerful server.

Ready to Build Your Pi Agent?

Deploy OpenClaw on your Raspberry Pi and start building today.

Get Started →

How to Run OpenClaw on Raspberry Pi