The real leap was going from gpt-4 to sonnet 3.5. 4o was meh, o1 was barely better than sonnet and slow as hell in comparison.

The native voice mode of 4o is still interesting and not very deeply explored though imo. I'd love to build a Chinese teaching app that actual can critique tones etc but it isn't good enough for that.

Its strange how Claude achieves similar performance without reasoning tokens.

Did you try advanced voice mode? Apparently it got a big upgrade during gpt 5 release - it may solve what you are looking for.

Yeah, I'd love something where you pronounce a word and it critiques your pronunciation in detail. Maybe it could give you little exercises for each sound, critiquing it, guiding you to doing it well.

If I were any good at ML I'd make it myself.