Compared to manually installing Node.js, configuring the environment, and deploying step by step, Tencent Cloud's application image approach eliminates all of that. When purchasing a server, simply select the OpenClaw image—after boot, the environment and software are already installed and ready. You go straight to the console for configuration. For newcomers, this is the fastest path to getting started.
1. What you need before starting
Three things: a Tencent Cloud account, a cloud server, and an API key for your AI model.
The API key can be prepared before buying the server. OpenRouter is the recommended choice—one key connects to over 400 models without needing separate registrations for each platform. Go to openrouter.ai, create an account, generate a key on the API Keys page, and keep it somewhere handy.
2. Purchase the server
Log into the Tencent Cloud console, navigate to Lighthouse Application Server, and click Create Instance.
| Setting | Recommended choice |
|---|---|
| Image type | Application image |
| Image | OpenClaw |
| CPU / Memory | 2 cores 2GB (personal use) / 2 cores 4GB (stable operation) |
| Region | Singapore / Silicon Valley / Tokyo |
| Bandwidth | 3Mbps or above |
Choose Singapore, Silicon Valley, or Tokyo for the region. These nodes offer low latency and reliable connectivity to overseas APIs like OpenAI, Claude, and OpenRouter, with minimal timeout risk. Domestic nodes tend to cause problems when calling overseas APIs and aren't recommended for this use case.
Pricing reference: 2-core 2GB runs approximately ¥30–60/month; 2-core 4GB is roughly ¥60–100/month. New user discounts are usually available.
Click Buy Now and complete payment. The server typically activates within 1–2 minutes, and the public IP will appear in your instance list.
3. Access the OpenClaw console
Once the server is running, OpenClaw starts automatically—no additional steps required.
Open a browser and go to:
http://your_server_IP:18789If the page doesn't load, check the firewall settings in the Tencent Cloud console.
4. Open the firewall port
In the Lighthouse console, find Firewall settings and add an inbound rule:
Protocol: TCP
Port: 18789
Source: 0.0.0.0/0Save the rule, then revisit the console address—it should load normally.
5. Configure the AI model
In the OpenClaw backend, find the model configuration page and fill in:
API Base URL: https://openrouter.ai/api/v1
API Key: your OpenRouter key
Default model: minimax/minimax-01Setting a cost-effective entry-level model as the default is a good practice—use it for routine tasks and switch to a more capable model only when genuinely needed. This keeps API costs low.
After saving, click Test Connection to confirm the API is working.
6. Connect a messaging channel
OpenClaw supports receiving instructions and push notifications via Telegram, Feishu, DingTalk, and WeCom. Using Telegram as an example:
Search for BotFather in Telegram and send /newbot. Follow the prompts to name your bot. You'll receive a token in this format:
1234567890:ABCdefGHIjklMNOpqrSTUvwxYZGo back to the OpenClaw backend, find Telegram under channel configuration, paste the token, and save. Find your new bot and send /start—a reply confirms the connection is working. You can now send instructions to the AI directly from Telegram on your phone.
7. Confirm automatic restart
Application images are typically configured to start on boot, but it's worth verifying. Log into the server (via the console's Login button or local SSH) and check the service status:
systemctl status openclawIf you see active (running), the service is healthy and auto-start is configured.
If the service isn't running, start and enable it manually:
systemctl start openclaw
systemctl enable openclaw8. Common issues
If the backend is unreachable, check in this order: whether the firewall rule opens port 18789, whether the service is running (systemctl status openclaw), and whether the IP address is correct.
If the service slows down or crashes after running for a while, insufficient memory is usually the cause. Add a swap file to ease the pressure:
fallocate -l 2G /swapfile
chmod 600 /swapfile
mkswap /swapfile
swapon /swapfile
echo '/swapfile none swap sw 0 0' >> /etc/fstabTo update OpenClaw to the latest version:
npm update -g openclaw
systemctl restart openclawTo view live logs for troubleshooting:
journalctl -u openclaw -fSummary
The entire Tencent Cloud one-click deployment process comes down to three steps: select the OpenClaw application image and purchase a server, open the firewall port, and configure the API in the console. No command-line work required, and the whole thing can be done in under five minutes. Paired with OpenRouter's multi-model routing, this is currently the fastest and most cost-effective way for users in China to get a 24-hour AI assistant up and running.