> if you’re willing to say that some models that outperform GPT-4 on some benchmarks count as being GPT-4 level How else woild you degine being on the same lev

> if you’re willing to say that some models that outperform GPT-4 on some benchmarks count as being GPT-4 level How else woild you degine being on the same level? You are better at some things and worse at others.