Send one prompt to GPT-4o-mini and Claude Haiku. Compare side-by-side with XEB scoring.
Try an example