Local (Ollama / LM Studio)

Run models entirely on your own machine — no API key, no data leaving your network.

Setup

1. Start your local server (e.g. ollama serve, or LM Studio's local server mode)
2. Open Settings → enable the Local provider card, and set the endpoint URL (defaults typically match Ollama's http://localhost:11434 or LM Studio's OpenAI-compatible port)
3. Click Test Connection to confirm

Model selection

The model list is fetched from your local server's model listing endpoint.

Token usage

Exact input/output counts are returned by the local server's OpenAI-compatible API response.

Privacy

Nothing you generate through the Local provider ever leaves your machine — useful if your product briefs or designs are sensitive.