Configure the model on the right, write a prompt below, and run it.
Responses are simulated - no live inference.