Model Connections

XClaw is model-agnostic. It doesn't lock you into any single vendor — you can connect multiple models simultaneously and use whichever fits the task.

This design comes from a practical observation: no single model is best at everything. Claude excels at long-context reasoning and code comprehension, GPT-4o is stronger in certain creative tasks, and local models are irreplaceable when data privacy matters. XClaw lets you avoid either-or choices between them.

Authentication depends on the provider: API keys (Anthropic, OpenAI, Google), OAuth sign-in (Claude Pro/Max, ChatGPT Plus), cloud credentials (AWS Bedrock, GCP Vertex), or fully local (Ollama). XClaw also supports any OpenAI-compatible API endpoint, including aggregation services like OpenRouter.

Model switching happens at two levels. The workspace level sets a default affecting all new sessions. The session level can override it at any time — you can even switch models mid-conversation.

All credentials are encrypted and stored locally. Token refresh is handled automatically by XClaw — you don't need to manually manage expired OAuth tokens.

How to

Open Settings > AI Models and click "Add Connection" to get started. You can paste an API key directly or use OAuth for one-click sign-in. Once connected, the model appears in your list and you can test the connection status anytime.

Want to set a workspace default model? In the AI Models list, click the star icon next to a model to make it the default for that workspace. All newly created sessions will use it.

Switching models mid-session is easy too — click the model selector above the session input box and pick a different one. This only affects the current session, not the workspace default.

If you're using Ollama for local models, make sure the Ollama service is running locally — XClaw will automatically discover the available models.