Configure a model

Add a DeepSeek or compatible model provider before starting your first task.

Harness needs at least one model provider before it can answer in a session. The fastest path is to add a DeepSeek API key from the embedded Harness settings.

Add DeepSeek

1. Open model settings

   In the Harness sidebar, open Settings → Models. This page belongs to the embedded Harness interface, not the desktop shell’s Configuration dialog.

2. Enter your API key

   Find the DeepSeek provider card, enter a key from the DeepSeek Platform, and save it.

3. Select a model

   Return to the session composer and choose an available model. The selection becomes the default for new sessions.

4. Send a small test task

   Select a workspace and ask Harness to summarize the repository. Model changes apply to the next request; the Harness service does not need to restart.

Never paste an API key into a chat message, repository file, screenshot, or issue. Harness stores provider credentials separately and returns only a redacted descriptor to the browser UI.

Other providers

Choose Add provider for an installed provider such as Anthropic or OpenAI. Choose Add custom provider for an OpenAI-compatible company gateway or self-hosted endpoint. Custom providers require a stable lowercase provider ID, base URL, protocol, credential, and at least one model.

Provider setup is a Harness feature and can evolve independently from the desktop shell. For advanced routing, image input, and compatibility options, consult the upstream model guide.

Next, learn the desktop and Harness interface.