OpenClaw Cloud Hosting — Managed OpenClaw VPS

ClawCloud gives you a fully configured cloud VPS with a dedicated API key and 94 curated AI models included. No SSH, no API key setup, no maintenance. Deploy in under a minute.

Compare plan tiers on the OpenClaw pricing page.

Prefer to use your own API key? See BYOK VPS mode.

Telegram Bot Hosting Discord Bot Hosting Feishu Bot Hosting

What OpenClaw needs to run

The minimum requirements before you can even start the gateway.

Ubuntu 24.04 LTS

Long-term support with security patches through 2029

Node.js 22

Required by OpenClaw's gateway and plugin system

2+ GB RAM + swap

The gateway, LLM streaming, and tool execution run concurrently in memory

UFW firewall

Only SSH (22), HTTP (80), HTTPS (443), and the agent port (8080) are open

Stable network

WebSocket connections to Discord/Telegram/Feishu need a persistent, low-latency link

If you self-host, you configure all of this yourself. The OpenClaw docs cover installation, but the docs assume you already have a server ready. Most people get stuck between "I have a VPS" and "OpenClaw is actually responding to messages."

What a production setup adds

Getting OpenClaw installed is one thing. Keeping it running is another.

Dedicated openclaw user

OpenClaw runs under its own non-root user with loginctl linger enabled, so the gateway survives SSH disconnects

User-level systemd service

The gateway runs as a user-level systemd unit. It starts on boot, restarts on crash, and logs to journalctl

Health monitoring agent

A separate process on port 8080 reports gateway status, config state, and version info

Config hot reload

OpenClaw watches its config file and applies safe changes (models, channels, routing) without restarting

Without this, your bot stops when you close your terminal, crashes without restarting, and runs as root with full system access. Reddit is full of posts from people stuck at this stage.

What ClawCloud bundles on every instance

Each deployment is a dedicated VM built from a pre-configured snapshot.

Software	Version	Why it's included
OpenClaw	Latest	The AI assistant gateway itself
ClawHub	Latest	Tool and skill marketplace for OpenClaw plugins
Node.js 22	22.x LTS	Runtime for the gateway and agent
Chromium	System	Headless browser for web scraping tools
Python 3 + pip	System	Required by data science and automation tools
Build tools	gcc, g++, make	Native npm modules that compile on install
Utilities	git, jq, unzip, zip	Dependency for various OpenClaw skills

The snapshot also includes 1 GB of swap, a configured firewall (UFW), and the openclaw system user with proper permissions. New droplets boot from this snapshot, then cloud-init writes instance-specific config (API keys, channel tokens, model settings) and starts the gateway.

Need to decide whether to rebuild an existing machine? Read Destroy vs Regenerate: When to Rebuild Your OpenClaw Server.

What you don't have to think about

ClawCloud handles the operational layer so the bot just works.

API key provisioning

Managed-mode instances get a dedicated API key with a per-plan credit limit. No key setup required.

Credit tracking and limits

Usage is polled every 5 minutes. When credits run low, switch to a free model from your dashboard or chat.

Health checks and auto-recovery

The dashboard monitors each instance. If the gateway goes down or a key becomes invalid, you see it immediately.

94 curated quality models, switchable in chat

Every instance has access to models from Anthropic, OpenAI, Google, and 14 other providers. Switch with /model.

One-command channel setup

Paste your bot token in the deploy wizard. DM access, group policies, and permissions are configured automatically.

Automatic updates

Agent and configuration updates are pushed to live instances without downtime or manual SSH.

BYOK (bring your own key) is supported if you want to use your own API keys from Anthropic, OpenAI, or Google directly. In that mode, you skip the credit system entirely and pay your provider directly. See the getting started guide for a walkthrough.