Use Poolside models in OpenClaw

​

Watch the setup video

​

Prerequisites

• You have OpenClaw installed. For installation options, see Install OpenClaw.
• You have one of the following:
  • A Poolside Platform API key. To create one, go to platform.poolside.ai.
  • The base URL and API key, if required, for your self-managed model endpoint.

​

Steps

1. Install the Poolside provider plugin from ClawHub.
   openclaw plugins install clawhub:@poolside/openclaw-provider
   
2. Authenticate with Poolside.
   • Poolside Platform
   • Self-managed Poolside inference

   openclaw onboard --classic
   

   Review and accept the security disclaimer, select the default quick-start setup mode, then select Poolside when prompted for your primary authentication and model provider. Paste your Poolside Platform API key when prompted.

   If you run Poolside model inference on your own infrastructure, use the model’s OpenAI-compatible endpoint at https://<model-hostname>/v1. Use the API key configured for the endpoint, if it requires one.

   • If your model endpoint requires an API key, add it to ~/.openclaw/.env on the Gateway host:
     POOLSIDE_API_KEY=<api-key>
     
   • Set the provider’s base URL to the model’s OpenAI-compatible endpoint:
     openclaw config set models.providers.poolside.baseUrl "https://<model-hostname>/v1"
     
   • Restart the Gateway to load the API key and base URL:
     openclaw gateway restart
     
3. Set a default model.
   openclaw models set <poolside-model-id>
   
   The plugin registers the available Poolside models. To find model IDs for your access method, see List available models. To switch models for one chat without changing the default, use the /model command in the chat.
4. Send your first prompt. Start an agent session or send a message on a connected channel. Use a prompt that is specific and has a clear expected result, such as one of these examples:
   Every morning at 08:00, check the 10 trending papers on Hugging Face (sorted by upvotes).
   - For each paper, show: title, upvotes, GitHub repo (if any), and a 1-line AI summary.
   - Notify me when you notice a spike in upvotes for a given paper.
   
   Ask me five questions to get to know how I work, then update your USER file with what you learn.
   
   Confirm OpenClaw responds using the selected Poolside model. To check the primary model and provider authentication directly, run openclaw models status.

​

Related resources

• Coding agents
• Poolside API
• Supported models
• OpenClaw documentation