When you deploy an OpenClaw bot on ClawCloud, you don't need an API key. The managed option — which is the default — handles all AI provider setup for you. You pick a model, deploy, and start chatting. This guide explains what that means and how it works.
What is managed mode?
Managed mode means ClawCloud provides the AI connection for your bot. When you select Claude, GPT, or Gemini during deployment, ClawCloud provisions a dedicated API key behind the scenes and routes your bot's requests through OpenRouter. You never see or manage this key.
The managed credits addon gives your instance a monthly pool of AI credits that covers the API usage. The bot is fully functional from the moment it comes online.
How it works
- You pick a model during the 3-step deploy wizard (Claude, GPT, or Gemini)
- ClawCloud creates a dedicated API key linked to your instance — routed through OpenRouter
- Your bot starts responding using that model immediately after deployment
- Usage draws from your credit pool — tracked automatically in the dashboard
You don't create accounts with OpenAI, Anthropic, or Google. You don't paste API keys. You don't configure billing with an AI provider. ClawCloud handles all of it.
What you get
Your base plan covers the server, hosting, and security updates. ClawCloud plans start at $29–$109/month for the server and support. Add the managed AI credits addon for hands-off AI billing — no separate AI provider account needed. See the hosting breakdown for what's installed on each instance.
Credits reset monthly. Each billing cycle, your credit pool resets to the full amount for your addon tier:
| Addon Tier | Monthly Price | AI Credit Budget |
|---|---|---|
| Small | +$9/mo | $8 |
| Medium | +$28/mo | $25 |
| Large | +$66/mo | $60 |
| XLarge | +$110/mo | $100 |
Model switching is free. Change your bot's AI model anytime with the /model command in chat. Switching between GPT, Claude, and Gemini doesn't cost extra — all usage comes from the same credit pool.
Manual fallback when credits run out. If your bot uses all its credits before the billing cycle ends, it stops responding on paid models. If Backup Models is enabled, you can switch to a supported :free model manually with the /model command to keep your bot running until credits reset. See the credits guide for details on switching and the free model fallback guide for step-by-step instructions.
Default models
ClawCloud picks cost-efficient default models to give you the best value from your credit pool:
| Provider choice | Default model |
|---|---|
| GPT | GPT-4.1 Mini |
| Gemini | Gemini 2.5 Flash |
| Claude | Claude Haiku 4.5 |
These models deliver strong quality at a lower per-message cost. GPT-4.1 Mini handles ~11,400 messages per month on the Lite plan, compared to ~2,350 on the full GPT-4.1.
These defaults only set the starting model. You can switch to any of the 100+ curated models after deployment using /model. See the model switching guide for the full list.
Managed vs. BYOK
ClawCloud also supports BYOK (bring your own key) for users who want direct control over their AI provider. Here's how they compare:
| Managed (addon) | BYOK (default) | |
|---|---|---|
| API key | Provided by ClawCloud | You provide your own |
| Setup | Add credits addon during deploy | Create provider account, generate key, paste during deploy |
| Billing | Addon price on top of base plan | Separate bill from your AI provider |
| Credit limits | $8–$100/month depending on addon tier | No limits from ClawCloud |
| Model access | 100+ curated models via OpenRouter | Models available from your provider |
| Dashboard tracking | Credit usage bar, usage warnings | No usage tracking |
| Best for | Quick setup, no provider accounts | Heavy usage, specific provider needs |
You choose between managed and BYOK during deployment. The setting applies to the instance for its lifetime.
Who should use managed mode
Managed mode works well when you:
- Want to deploy fast without creating AI provider accounts
- Prefer a single bill for hosting and AI usage
- Run a bot with moderate traffic (under a few thousand messages per month)
- Don't want to manage API keys and provider billing
If your bot handles very high message volume or you need a model not available through OpenRouter, BYOK gives you that flexibility.
Getting started
The managed option is available as an addon in the deploy wizard. Pick your plan, add the managed credits tier, connect your channel, choose an AI model, and deploy.
Troubleshooting
Bot not responding after deployment — Give it 30–60 seconds. Check the dashboard for provisioning status. If it stays in "deploying" longer than 2 minutes, contact help@clawcloud.sh.
Credits ran out and bot stopped — If Backup Models is enabled, switch to a supported :free model using /model in the bot. See Keeping Your OpenClaw Bot Running When Credits Run Low.
Want to switch from managed to BYOK — This requires re-provisioning your instance. Use the dashboard Reconfigure flow. See Destroy vs Regenerate for details.