Select models in the sidebar, type a prompt, and all models will respond simultaneously.