Yes what I’ve learned is that “prompt engineering” is a lie.

Once you’ve got a great prompt that’s all that matters. Next is just how different models answer the same prompt