Settings
Connect an OpenAI-compatible API to run evaluations against real models. The key is kept server-side and never returned to the browser.
No LLM — simulator mode
Evaluation runs use a deterministic simulator. Connect a key below for real model calls.