Ampere
Ampere.sh
Installation Guide

How to RunOpenClaw on Raspberry Pi

Set up OpenClaw on a Raspberry Pi with simple steps, system requirements, and limitations.

8 min read
Mar 18, 2026
Ampere Team

Running OpenClaw on a Raspberry Pi gives you a personal AI agent that stays on 24/7 — without a cloud bill.

If you prefer skipping manual setup, you can also deploy it instantly using Ampere.sh.

Can a Raspberry Pi Actually Run OpenClaw?

Yes — with the right model. OpenClaw is a Node.js application, and Node runs well on the Pi's ARM64 chip. The limiting factor isn't the CPU — it's RAM.

OpenClaw's gateway process typically uses 150–300MB of memory at idle, and spikes higher during active conversations. That's fine on a 4GB or 8GB Pi. On a 1GB Pi 3, you're fighting for every megabyte.

Not Recommended

Pi 3 (1GB)

Possible with swap, but painful. Not recommended.

Tight

Pi 4 (4GB)

Works well for a single agent. The sweet spot for most users.

Recommended

Pi 5 (8GB)

Comfortable. Multiple agents, Docker sandbox, room to breathe.

Hardware & Prerequisites

DeviceRaspberry Pi 4 or 5
RAMMinimum 4GB (8GB recommended)
OSRaspberry Pi OS (64-bit)
Storage16GB+ free (SSD recommended for better speed)
DependenciesNode.js v22+, curl, git, build-essential
NetworkStable internet

Before running the installer, make sure the essential build tools are present:

sudo apt-get update sudo apt-get install -y curl git build-essential

If this setup feels a bit resource-heavy for your device, you can run OpenClaw in about a minute on Ampere.sh without configuring anything locally — no hardware, no dependencies.

Step-by-Step Installation Guide

Follow these steps in order. Each one accounts for the Pi's ARM architecture — don't skip the OS check or the Node install even if you think you already have them set up.

1

Check Your OS is 64-bit

OpenClaw requires Node.js 22+, which only runs on 64-bit OS. Many Pi setups still default to 32-bit. Check yours first:

uname -m

If you see aarch64 you're good. If you see armv7l, re-flash with Raspberry Pi OS (64-bit) using the Raspberry Pi Imager before continuing.

Installing Raspberry Pi OS (64-bit) using the official imager
Installing Raspberry Pi OS (64-bit) — use the official Raspberry Pi Imager
2

Install Node.js 22+

Don't use the version from Raspberry Pi OS's default repos — it ships an outdated Node. Install directly from NodeSource:

curl -fsSL https://deb.nodesource.com/setup_22.x | sudo -E bash - sudo apt-get install -y nodejs

Verify: node -v should show v22.x.x or newer.

NodeSource DEB repository — install Node.js 22 on Raspberry Pi
NodeSource repository — install Node.js 22 LTS on Raspberry Pi OS
3

Run the OpenClaw Installer

Run the official install script — it detects your Node version, installs the CLI globally, and launches the onboarding wizard:

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

If openclaw is not found after install, run exec $SHELL -l to reload your shell session.

Running the OpenClaw installer on Raspberry Pi terminal
OpenClaw installer running on Raspberry Pi — detects Node v22 and installs successfully
4

Complete Onboarding

The onboarding wizard will walk you through:

  • Choosing an AI provider (Anthropic, OpenAI, etc.) and entering your API key
  • Connecting a messaging channel — Discord, Telegram, or WhatsApp
  • Setting a gateway token for dashboard access

After it completes, verify with: openclaw doctor

OpenClaw first-run onboarding and security prompt
OpenClaw first-run onboarding — security setup and AI provider configuration
5

Start the Gateway & Enable Auto-Start

Start the gateway daemon and install it as a system service so it restarts automatically after every reboot:

openclaw gateway start openclaw gateway install

Check it's running: openclaw gateway status. Your Pi is now a 24/7 AI agent host — no cloud required.

openclaw gateway status — systemd service running on Raspberry Pi
openclaw gateway status — service enabled and running, bound to port 18789

Troubleshooting

openclaw: command not found

Run exec $SHELL -l and try again. If still missing, rerun the install script.

exec $SHELL -l

Node version too old

Install Node 22+ from NodeSource and re-run onboarding.

curl -fsSL https://deb.nodesource.com/setup_22.x | sudo -E bash - sudo apt-get install -y nodejs openclaw onboard

Control UI not reachable from your laptop

Confirm you're using the Pi's local IP address (not localhost) and that both devices are on the same network.

# Open in your browser: http://<PI_IP>:18789/

Gateway not running after onboarding

Run the status command and check service logs.

openclaw gateway status openclaw gateway start

Channel added but messages don't arrive

Re-check your bot token and channel config; re-add the channel if needed.

openclaw channels list openclaw channels login

If you don't want to deal with these issues, you can deploy OpenClaw on a managed hosting platform with:

  • ✓ No setup
  • ✓ No debugging
  • ✓ No environment configuration

Which Pi Should You Use?

Not all Pis are created equal. Here's how different models handle OpenClaw:

Pi 3B+ (1GB)Pi 4 (2GB)Pi 4 (4GB)Pi 4 (8GB)Pi 5 (8GB)
Installs?With swapYesYesYesYes
Runs reliably?NoMostlyYesYesYes
Docker sandbox?NoNoLimitedYesYes
Multi-agent?NoNoNoPossibleYes
RecommendedNoBudgetBest valuePower userMax perf

Frequently Asked Questions

Does OpenClaw support ARM processors?
Yes. OpenClaw runs on Node.js, which fully supports aarch64 (64-bit ARM). That's the architecture used by Raspberry Pi 4, Pi 5, and most modern ARM single-board computers.
Can I run OpenClaw on a Raspberry Pi 3?
It's technically possible on the Pi 3B+ (which has 1GB RAM), but you'll be running at the absolute edge. You'll need a swap file, can't run much else alongside it, and performance will be noticeably slower. Pi 4 with 4GB is the realistic minimum for a good experience.
Does OpenClaw work with Docker on Raspberry Pi?
It can, but Docker adds significant memory overhead. On a Pi 4 with 4GB, running OpenClaw natively (without Docker) is the better choice. If you have an 8GB Pi 4 or a Pi 5, Docker becomes more viable. The image build alone requires about 2GB of free RAM.
Can I connect WhatsApp to OpenClaw on my Pi?
Yes — run 'openclaw channels login' after setup and scan the QR code with your phone. It works the same way regardless of what hardware you're running on.
What happens if my Pi loses power?
If you've set up OpenClaw as a system service (Step 5 in this guide), it will automatically restart when power returns. Your conversation history and memory files are preserved on disk.
Is the SD card fast enough for OpenClaw?
It works, but SD cards are the biggest performance bottleneck on most Pi setups. OpenClaw reads and writes session data, logs, and memory files frequently. If you notice lag, switching to a USB SSD boot drive makes a dramatic difference — often 5-10x faster for disk operations.
Can I run multiple AI agents on one Raspberry Pi?
On a Pi 5 with 8GB RAM, you can comfortably run 2-3 agents. On a 4GB Pi 4, stick to one agent and keep background services minimal. Each agent consumes memory for its Node.js process and conversation context.
Does the Pi need to be connected to the internet all the time?
Yes. OpenClaw needs internet access to reach AI model providers (like Anthropic, OpenAI, etc.) and to maintain connections with messaging platforms. Without internet, the gateway will start but won't be able to process messages.

Your agent deserves to run 24/7

Whether it's on a Raspberry Pi or in the cloud — OpenClaw keeps your AI agent always on. Try Ampere.sh for instant deployment with free credits.

Deploy on Ampere.sh →