OpenClaw Bot Not Responding

Before diving into details, run through these quick checks. Most issues are solved by one of these:

• Is the gateway running? → Run openclaw gateway status
• Is the API key set? → Check your model provider API key in settings
• Is the API key valid? → Test it directly with your provider
• Is the channel connected? → Check WhatsApp/Telegram/Discord/Slack connection status
• Is the server online? → SSH into your VPS or check your hosting dashboard
• Is the disk full? → Run df -h on your server
• Did you update recently? → Updates can break configs

The most common reason OpenClaw stops responding is a missing or invalid AI model API key. Without a working key, the agent cannot call any AI model and silently fails.

Look for warnings about missing model configuration or API key errors in the output.

• Check your API key - make sure it is set in your OpenClaw configuration or environment variables
• Verify the key is valid - go to your provider dashboard (OpenAI, Anthropic, Google) and check the key status
• Check for expired keys - some providers expire keys after a set period
• Check your credits - if you ran out of API credits, the provider will reject requests
• Re-enter the key - if in doubt, generate a new key and update your config
• Restart the gateway - after changing the key, restart with openclaw gateway restart

For help choosing the right model and key, see the AI model guide and model change guide.

If the gateway process is stopped, nothing works - no messages, no workflows, no responses.

• Start the gateway: openclaw gateway start
• Restart if stuck: openclaw gateway restart
• Check logs for crash reason: look for OOM kills, port conflicts, or config errors
• Check if the port is in use: another process may be blocking the gateway port
• Set up auto-restart: use systemd, Docker restart policy, or a process manager so the gateway comes back after crashes

For detailed gateway setup, see the uptime and reliability guide.

The gateway is running and the API key is valid, but the bot does not respond on your messaging app. The channel connection probably dropped.

If you are self-hosting OpenClaw, server issues can bring everything down:

• Out of memory (OOM): Your server ran out of RAM. Check with free -h. Upgrade your VPS or reduce browser automation usage.
• Disk full: Logs, cache, and memory files filled the disk. Check with df -h. Clean old logs and temp files.
• Server rebooted: Kernel update, power outage, or provider maintenance. Set up auto-start so OpenClaw comes back automatically.
• Docker container stopped: Check with docker ps -a. Restart with docker start openclaw. Add restart: always to your compose file.
• SSL expired: Let's Encrypt certificates expire every 90 days. Check and renew with certbot.
• Port conflict: Another service is using the gateway port. Check with lsof -i :PORT.

For Docker setup, see the Docker guide. For VPS setup, see VPS installation.

Sometimes the problem is not OpenClaw - it is the AI provider. Your key is valid but the provider is having issues:

• Rate limits: You hit the API rate limit. Wait and retry, or use a different model. See API cost guide.
• Provider outage: OpenAI, Anthropic, and Google all have occasional outages. Check their status pages.
• Model unavailable: The specific model you selected may be deprecated or temporarily unavailable. Try switching models.
• Insufficient credits: Your API account ran out of credits. Top up your balance.
• Region restrictions: Some API providers restrict access from certain regions. Check if your server location is supported.

Most of these problems exist because of self-hosting complexity - server management, Docker configuration, port setup, SSL, monitoring, and manual restarts.

Managed hosting on Ampere.sh handles all of this. The gateway runs automatically, channels stay connected, API keys are configured through a dashboard, and if something breaks, it recovers without you touching a terminal.

• No gateway management - it runs and restarts automatically
• No server maintenance - updates, backups, and monitoring are handled
• No port or firewall issues - networking is pre-configured
• API key setup through dashboard - no environment variables to manage
• Channel connections managed through UI - no manual reconnection

If none of the above fixed your issue:

• Check OpenClaw docs: The official documentation has detailed troubleshooting for specific setups
• Join the community: The OpenClaw Discord and GitHub discussions have active users who can help
• Check logs carefully: The answer is almost always in the logs - look for the first error, not the last one
• Try a clean restart: Sometimes stopping everything, clearing cache, and starting fresh fixes mysterious issues
• Update OpenClaw: If you are on an old version, the bug may already be fixed. See the update guide.