Comment by Flow

These new models consume so many tokens. I’m very satisfied with GPT-5.2 on High. I hope they keep that one for many years