Feels like a really good engineering in a wrong direction. Who said that the audio is good interface anyway? Audio is hard to edit, slow and has low-information density. If I want to talk to someone and have low-information but pleasant exchange I can just to talk to real people, I don't need computers for it.
I guess it is useful for some casual uses, but I really wish there was more focus on the reasoning and intelligence of the model itself.
It does not mean that it is objectively many times better than GPT. It can be a disagreement of interpretation, Ilya and co strongly believing that it is not safe to release, while Sam thinking everything is fine. The capabilities themselves are not very relevant.
I guess it is useful for some casual uses, but I really wish there was more focus on the reasoning and intelligence of the model itself.