Send a prompt and see which model Flow AI picked, the latency, and exactly what it cost. Runs on a shared demo key (256-token cap).