Pick 2–6 models, write one prompt, run them all in parallel and see the results side-by-side. The same model can be added more than once with different settings.
