← Back to AI Hub

🚀 OpenClaw Agent Setup

System Requirements

OpenClaw runs on Linux (arm64, amd64, armv7) and macOS. Minimum:

• CPU: Dual-core 1.5+ GHz (single-board computers to servers)
• RAM: 2GB minimum, 4GB recommended
• Disk: 8GB free (more for models)
• Network: Stable internet connection

Installation

Linux (ARM64 / Raspberry Pi 5, Orange Pi)

# Update system
sudo apt update && sudo apt upgrade -y

# Install Node.js 22+
curl -fsSL https://deb.nodesource.com/setup_22.x | sudo -E bash -
sudo apt install -y nodejs

# Install OpenClaw
npm install -g openclaw

# Initialize agent
openclaw init --non-interactive
openclaw pairing approve telegram CODE_HERE

# Start
openclaw agent run

Linux (AMD64 / x86 desktop, server)

# Update system
sudo apt update && sudo apt upgrade -y

# Install Node.js 22+
curl -fsSL https://deb.nodesource.com/setup_22.x | sudo -E bash -
sudo apt install -y nodejs

# Install OpenClaw
npm install -g openclaw

# Initialize agent
openclaw init --non-interactive
openclaw pairing approve telegram CODE_HERE

# Start
openclaw agent run

macOS

# Install Homebrew if needed
/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

# Install Node.js
brew install node@22

# Install OpenClaw
npm install -g openclaw

# Initialize agent
openclaw init --non-interactive
openclaw pairing approve telegram CODE_HERE

# Start
openclaw agent run

Configuration

Edit ~/.openclaw/openclaw.json:

{
  "agent": {
    "name": "MyAgent",
    "model": "anthropic/claude-sonnet-4-5"
  },
  "channels": {
    "telegram": { "enabled": true }
  },
  "plugins": {
    "entries": {
      "telegram": { "enabled": true }
    }
  }
}

Model Selection & Local Inference

OpenClaw can use both cloud APIs (Anthropic Claude, OpenAI) and local models (Ollama, vLLM).

Cloud Models (Default)

• Anthropic Claude Sonnet 4.5: Primary model, best reasoning
• CKBDev API: Free shared tier (fallback)
• OpenAI GPT-5: Optional, requires API key

Local Models

For cost savings (85% reduction), run local Ollama inference:

# Install Ollama
curl https://ollama.ai/install.sh | sh

# Download a model
ollama pull qwen2.5:7b

# OpenClaw will auto-detect on localhost:11434

Hardware & Model Recommendations

See Local LLM Inference Guide for 20+ hardware tiers (Raspberry Pi to H100) with:

• ✅ Complete hardware specs (Jetson, Mac Studio, RTX 3090/4090/5090, etc.)
• ✅ Cost estimates ($15 to $40k+)
• ✅ Throughput benchmarks
• ✅ Model recommendations per tier

Configure Local Inference in OpenClaw

Edit openclaw.json:

{
  "agent": {
    "name": "MyAgent",
    "model": "ollama/qwen2.5:7b"  ← Switch to local model
  },
  "providers": {
    "ollama": {
      "endpoint": "http://localhost:11434",
      "fallback": "anthropic/claude-sonnet-4-5"  ← Cloud fallback if Ollama down
    }
  }
}

Local Model Setup (Optional)

For cost savings, use local Ollama inference:

# Install Ollama
curl https://ollama.ai/install.sh | sh

# Run locally
ollama run qwen2.5:7b

# Configure OpenClaw to use local inference
# See: Local LLM Inference guide in AI Hub

Troubleshooting

• Telegram auth fails: DM the bot first, then use correct code
• Model timeouts: Increase timeout in config, check local inference health
• Disk full: Models can be large; use quantized versions (GGUF)
• Permission denied: You may need to add your user to the dialout group: sudo usermod -a -G dialout $USER

← Back to AI Hub