Local-first CLI for benchmarking LLMs on real hardware — quality, speed, reliability, and a real multi-turn agent loop.
By chatting or signing in you agree to the Terms and chat-message logging (revocable in History).