1
Sign in and connect
Sign in with Google, then connect your Telegram or Slack bot and your LLM provider (OpenAI, Anthropic, Gemini, or others).
Cloud Hosting
OpenClaw Cloud Hosting
Run OpenClaw without owning day-2 infrastructure work. You get managed operations, security defaults, and usage visibility while keeping control of model credentials and bot setup.
What it means
Cloud hosting for OpenClaw means your assistant runs on managed infrastructure with security-by-design defaults. You don't touch servers, networking, or update cycles. You bring your LLM provider credentials and messaging platform tokens, and the platform handles the rest.
What this page covers
This is a product-focused breakdown of runtime boundaries, security controls, and operating model. If you want Q&A style guidance first, start with Answer Hub. For the self-managed path, use the OpenClaw setup guide plus the troubleshooting checklist.
2 min
Time to first message with guided setup.
Zero infra
No servers to provision, patch, or monitor.
BYOK
Bring your own LLM keys. View hosting plans.
Cloud vs Self-hosted
Cloud hosting vs self-hosting for OpenClaw
| Factor | Self-hosted | Cloud hosting |
|---|---|---|
| Setup time | Hours to days | Minutes |
| Server management | You handle provisioning, OS, updates | Fully managed |
| Security hardening | You design and maintain | Built-in isolation, no public IP |
| Credential storage | You secure API keys in your runtime | Encrypted at rest (AES-256-GCM), keys separated from agent |
| Monitoring | You set up logging and alerts | Built-in usage and cost analytics |
| Access control | You implement allowlists | Allowlist-only messaging by default |
| Updates | You track and apply | Managed updates |
How it works
3 steps to deployed OpenClaw
2
Configure access
Set up your allowlist so only approved users can message your assistant. Credentials are encrypted and stored separately from your runtime.
3
Start chatting
Your instance deploys to an isolated environment with no public IP. Monitor usage and cost by model directly in the dashboard.
Real product screens
What this looks like in the actual app
Security + Cost Control
What you get with hosted OpenClaw
- Isolated runtime: your assistant runs in its own environment, separated from other users and internal services.
- No public IP exposure: your instance is not directly reachable from the public internet.
- Encrypted credentials: your LLM API keys and bot tokens are encrypted at rest with AES-256-GCM and never exposed to the runtime.
- Allowlist-only access: only people or groups you approve can message your bot on Telegram or Slack.
- Per-model cost tracking: see token usage and cost broken down by model and time period in the dashboard.
- Workspace export: export your workspace anytime. Secrets are excluded or redacted.
Cost control
With BYOK (bring your own keys), you pay your provider directly for model usage. Hosting plan pricing is shown in our pricing section. The dashboard shows token burn by model so you can catch surprises early.
Trust & operations
How this is operated in practice
Reliability + security
- Uptime target: 99.9% for managed hosting layer.
- Network posture: instances are not exposed with a public IP.
- Isolation: runtime boundaries separate user environments.
- Credential protection: AES-256-GCM encryption at rest; keys are kept separate from agent runtime.
- Access control: allowlist-only messaging to reduce unauthorized access risk.
Maintenance + support
- Update cadence: daily rollouts for security hardening, base OpenClaw updates, and stability improvements.
- Support channels: gregory@lemon-ai.com and @GregoryPotemkin.
- Response expectation: usually within 1 hour, and no later than 12 hours.
FAQ
Common questions about this managed option
You get a managed OpenClaw runtime with deployment, patching, and baseline security controls handled by the platform team.
Self-hosting gives deeper low-level control but requires setup, hardening, patching, and monitoring. Managed runtime removes that recurring work.
Isolated runtime boundaries, no public IP exposure, encrypted credentials (AES-256-GCM), and allowlist-based messaging access.
Yes. You can export your workspace from a self-hosted instance and import it into the managed platform. Secrets and tokens are excluded from exports for security.
Solo is $3.9/month for 1 active instance. Trio is $6.9/month for up to 3 active instances. For higher usage, contact the developer for a Custom plan. See pricing.
Ready
Deploy OpenClaw in 2 minutes.
No infrastructure work. Secure defaults. Clear cost visibility.