OpenClaw is an AI assistant that you deploy yourself on your own cloud server. Regular bots simply respond to messages, while this one can also act: send messages where needed, process data, work with files, and integrate with third-party services. You can connect to it via messengers (for example, Telegram), via API, or directly from the command line.
Creating a Server
To run OpenClaw, create a server with a suitable configuration through the control panel.
- Go to the «vStack Cloud» or «VMware Cloud» section.
- Go to the «Servers» section.
- In the personal account panel «Create a cloud server», click «Create server».
- Select the OS version, location, and configuration (you can choose a server with a ready-made configuration in the Fixed plan section), as well as the authorization method.
- In the «Applications» section, select the required application from the OpenClaw list.
- Come up with a server name and click «Create».
After creating the server, perform the initial setup of OpenClaw.
- Connect to the server via SSH from your computer.
- If you are using macOS or Linux, launch the terminal; on Windows - PowerShell.
- In the control panel, click on the server name and copy the string from the «SSH connection» block.
On the first connection, a request will appear to add the server to the trusted list:
Are you sure you want to continue connecting (yes/no/[fingerprint])?Type yes and press Enter. Enter the password from the «Root password» field on the «Connections» tab of the control panel.
Initial Configuration
To launch the OpenClaw setup wizard, call it manually with the command:
openclaw onboard
Read the disclaimer and agree to the terms by selecting «Yes». In the «Onboarding mode» section, select «QuickStart». At the «Model/auth provider» step:If you have an API key from an external provider, select it. If you plan to use the built-in AI agents, select «Skip for now».
Connecting a Custom Model via API
OpenClaw supports any model that implements the OpenAI-compatible (GPT-like) API format - including the Serverspace GPT API. This allows you to use your own model endpoint instead of external providers.
Step 1 - Get the API Key
Log in to your Serverspace account, go to the «GPT» section, and create a new API key. Copy the key value - you will need it in the next step.
Step 2 - Open the OpenClaw Configuration File
On the server, open the OpenClaw configuration file in a text editor:
sudo nano /etc/openclaw/config.yamlStep 3 - Specify the Custom Model Parameters
Find the model section and fill in the following fields:
model:
provider: openai-compatible
base_url: https://gpt.serverspace.ru/v1/chat/completions
api_key: YOUR_SERVERSPACE_API_KEY
model_name: gpt-4o- base_url - the base address of the Serverspace GPT API endpoint.
- api_key - the API key obtained in Step 1.
- model_name - the name of the model available in your Serverspace account (for example, gpt-4o or gpt-3.5-turbo).
Step 4 - Save and Restart OpenClaw
Save the changes (in nano: Ctrl+O, then Ctrl+X) and restart the service to apply the new settings:
sudo systemctl restart openclawStep 5 - Verify the Connection
Check that OpenClaw has successfully connected to the model:
openclaw statusIf the configuration is correct, the output will display the active model and the connected status. You can also send a test message via the command line:
openclaw chat "Hello, are you working?"A response from your custom model via the Serverspace GPT API confirms that the setup was completed successfully.
