Hardware Requirements

Choose the right Raspberry Pi for running OpenClaw.

Hardware Requirements

Running OpenClaw on a Raspberry Pi requires careful consideration of your hardware. Let’s look at what you need.

Minimum Requirements

To run OpenClaw, you’ll need:

• Raspberry Pi 4 (2GB RAM minimum) or Raspberry Pi 5
• Power Supply suitable for your model
• Networking (Ethernet or Wi-Fi)
• Storage (microSD card)
• Access (SSH or direct connection)

Recommended Specifications

For comfortable operation, we recommend:

Why More RAM Matters

OpenClaw runs several processes:

• Node.js runtime - ~100-150MB
• Gateway service - ~100-200MB
• Agent session - ~150-300MB
• Buffer for system - ~500MB

With 2GB, you’re at capacity. With 4GB or 8GB, you have breathing room for:

• Multiple agents
• Local model serving
• System processes
• Spikes in usage

Storage Considerations

Your microSD card needs space for:

• Raspberry Pi OS - ~3-4GB
• Docker installation - ~1-2GB
• OpenClaw container - ~500MB-2GB
• Your workspace data - 100MB-1GB+
• System updates - ~500MB
• Free space buffer - ~5GB minimum

Total recommended: 32GB card minimum

If you plan to:

• Run multiple models locally
• Store lots of data in your workspace
• Keep extensive backup copies

Consider a 64GB or 128GB card.

Power Supply

This is critical and often overlooked.

Using an undersized power supply causes:

• Mysterious crashes and reboots
• Corrupted filesystems
• Docker container crashes
• OpenClaw service failures

Buy a quality power supply. This is not the place to save money.

Networking

Ethernet recommended for:

• Stability (no Wi-Fi interference)
• Latency (important for real-time operations)
• 24/7 uptime scenarios
• Higher bandwidth

Wi-Fi acceptable if:

• You’re on 5GHz dual-band
• You’re close to the router
• It’s not mission-critical

Optional Hardware

AI HAT+

The official Raspberry Pi AI HAT+ is useful if you want to:

• Run local language models faster
• Experiment with on-device inference
• Learn about hardware acceleration

Not required, but nice for projects.

Case and Cooling

For continuous operation, consider:

• A good case (better airflow)
• Passive or active cooling
• Heatsinks on the SoC

OpenClaw isn’t computationally intensive, but the Pi might be doing other things. Cooling helps reliability.

UPS/Battery Backup

If you want 24/7 operation:

• A UPS (uninterruptible power supply) protects against power failures
• Prevents filesystem corruption
• Ensures graceful shutdowns

Choosing Your Pi

Pi 4 2GB - Minimum (Not Recommended)

Pros:

• Cheapest option
• Works technically

Cons:

• Very tight on memory
• Swaps to disk frequently (slow)
• Limited headroom for experimentation

Use case: Learning only, expect slowdowns

Pi 4 4GB - Entry Level (Acceptable)

Pros:

• Good balance of cost and performance
• Comfortable for basic OpenClaw usage
• Room to experiment

Cons:

• Limited if running multiple agents
• No room for local models

Use case: Single agent, cloud models only

Pi 4 8GB - Recommended

Pros:

• Plenty of headroom
• Can run local models
• Multiple agents simultaneously
• Future-proof

Cons:

• Costs more than 4GB

Use case: Serious experimentation, production use

Pi 5 8GB - Best

Pros:

• Faster CPU (A76 cores)
• 25% faster performance
• Thermal design improved
• Future-proof

Cons:

• Most expensive
• Might be overkill for basic use

Use case: Heavy workloads, production, local model serving

Sample Setups

Budget Setup ($50-70)

• Pi 4 4GB
• Basic power supply
• 32GB microSD
• Ethernet cable

✅ Good for learning ❌ Limited for production

Solid Setup ($100-130)

• Pi 4 8GB
• Quality 5A power supply
• 32GB microSD
• Case with cooling
• Ethernet

✅ Great for most use cases ✅ Room for growth

Premium Setup ($150-200)

• Pi 5 8GB
• Quality 5A power supply
• 64GB microSD
• Good case with active cooling
• Ethernet
• UPS battery backup

✅ Production-ready ✅ Runs anything OpenClaw can do

Before You Buy

1. Know what you’re using it for - Learning vs production changes your needs
2. Don’t cheap out on power supply - This causes more problems than undersized RAM
3. Get at least 4GB RAM - 2GB is too tight
4. Go 32GB+ storage - Cheaper than debugging storage issues
5. Consider cooling - Continuous operation benefits from it

Next Steps

Now that you know what hardware you need, let’s understand why Docker is the best way to run OpenClaw on it.

< Previous Next >

You can use the arrows  ← → on your keyboard to navigate between lessons.