Comment by bigupthewhole

Ive been using gpt 5.4 and 5.5 and honestly 5.4 is solving everything at the pace I need it. I'm the biggest bottle neck in terms of reviewing PRs and my own code. So having a model which can solve a complex task in 10 minutes vs 30 minutes doesn't really give me any meaningful improvement.

Also, the biggest factor is having a good planning phase. A good plan is better than even major model improvements.