Keys are stored locally in your browser. Calls go from your browser directly to the provider.
No configuration needed — the built-in coach runs fully offline using local heuristics and the built-in engine.
Load a model in LM Studio and start its local server. Pick one of the presets below, or type the exact model identifier LM Studio shows on its server page.
One key, 100+ models from Anthropic, OpenAI, Meta, Mistral, DeepSeek, xAI, Google, and more.
Zhipu AI's GLM models. The default endpoint is the China region — swap to a proxy if needed.