VPS Hosting — Run OpenClaw in the Cloud

VPS Hosting

Running OpenClaw in the cloud means your AI assistant is always online, even when your laptop is closed. The Gateway runs on the VPS and handles everything—you just connect when needed.

I moved my personal assistant to a $5/month VPS after getting tired of leaving my MacBook running 24/7. Best decision I made.

---

Quick Picks

Provider	Price	Best For
Railway	~$5/mo	One-click deploy, easiest setup
Northflank	~$5/mo	One-click, good free tier
Oracle Cloud	$0/mo	Free tier (ARM, capacity varies)
Fly.io	~$5/mo	Edge locations, good CLI
Hetzner	€4/mo	European, great value
GCP	Varies	Free tier available
AWS (EC2/Lightsail)	Varies	Free tier, most documented

---

Provider Guides

One-Click Deploys

• Railway → — Browser-based setup, no CLI needed
• Northflank → — Similar to Railway, good UI

Manual Setup

• Oracle Cloud → — Always Free tier (when available)
• Fly.io → — Good for edge deployment
• Hetzner → — Docker-based setup
• GCP → — Compute Engine setup
• AWS → — Video guide

---

How Cloud Setups Work

┌─────────────────┐     ┌──────────────────┐
│  Your Device    │     │      VPS         │
│  (laptop/phone) │────▶│    Gateway       │
│                 │     │  (source of truth)│
└─────────────────┘     └──────────────────┘
    Control UI              State + Workspace
    or Tailscale            backed up

Key points:

• Gateway runs on the VPS and owns state + workspace
• Connect via Control UI, SSH tunnel, or Tailscale
• Treat VPS as source of truth—back up regularly
• Keep Gateway on loopback by default; use gateway.auth.token if exposed

---

Security

{
  gateway: {
    bind: "127.0.0.1",     // Loopback only (safest)
    // OR
    bind: "lan",           // Local network
    auth: {
      token: "your-secret-token"  // Required if not loopback
    }
  }
}

Access methods (pick one):

• SSH tunnel: ssh -L 8000:localhost:8000 user@vps
• Tailscale Serve: Zero-config private access
• Direct exposure: Requires auth token

---

Using Nodes with VPS

Keep Gateway in the cloud, run nodes on local devices for:

• Screen/camera access
• Canvas rendering
• Local system.run capabilities

┌─────────────┐     ┌──────────────────┐
│ Local Node  │────▶│  VPS Gateway     │
│ (Mac/iOS/   │     │                  │
│  Android)   │     │  (brain in cloud)│
└─────────────┘     └──────────────────┘

See Nodes documentation for setup.

---

Minimum Requirements

• RAM: 512MB (1GB+ recommended)
• CPU: 1 vCPU (ARM works fine)
• Storage: 5GB+ for state and workspace
• Network: Stable connection for messaging platforms

---

Quick Deploy Checklist

1. Choose provider — Railway for easiest, Oracle for free
2. Deploy Gateway — Follow provider-specific guide
3. Set up access — SSH tunnel or Tailscale
4. Configure auth — Set gateway.auth.token if exposed
5. Set up backup — Rsync or cloud provider snapshots
6. Test messaging — Verify WhatsApp/Telegram works

---

Troubleshooting

Can’t connect to Gateway

Check:

1. Is Gateway running? ssh vps "pgrep -f openclaw"
2. Is port forwarded? Check your SSH command
3. Is firewall blocking? Check cloud provider rules

Messaging platforms not working

Check:

1. Are webhooks configured with VPS IP/domain?
2. Is the VPS reachable from the internet?
3. Check Gateway logs: openclaw logs --follow

High latency

Cause: VPS too far from messaging platform servers.

Fix: Choose a provider with regions closer to your users (Fly.io has good global coverage).

Still stuck? Our AI Setup Assistant can help with VPS setup issues.

---

What’s Next?

• Platforms Overview → — All supported platforms
• Gateway Remote → — Remote access details
• Nodes → — Pair local devices with cloud Gateway

---

Need help? Join the OpenClaw Discord or check the GitHub Issues.