I've actually done a lot of comparisons in this regard. 3.7 with the thinking tokens maxed out is about equal to o1-pro, occasionally o1-pro will be more elegan

I've actually done a lot of comparisons in this regard. 3.7 with the thinking tokens maxed out is about equal to o1-pro, occasionally o1-pro will be more elegant (in coding) but it's not a huge difference, for tasks like long context understanding with text summarization/aggregation I actually slightly prefer 3.7. That said, gemini-2.5-pro is waaay better than 3.7 at code and long context so its safe to assume it's better than o1-pro. If you want to do heavy coding right now go with Gemini.