Actually will agree that o1 pro was better than o3 at really deep bug finding/coding analysis. Which is also why i have the theory that they could just turn up

Actually will agree that o1 pro was better than o3 at really deep bug finding/coding analysis. Which is also why i have the theory that they could just turn up the compute to show better results, but dont do to cost. O3 and GPT5 seem heavily quantized, o1 pro was more raw Another thing I’ll add though, is o3 pro is better through the api than the chat website. They clearly constrain it unless you’re paying the absurd api cost