Audio2Text Transcription Quality

I really like the Quality of Voice Recognition, guess it depends on Input-types,

but for my case it works great. I also use Premiere 2025, and tried the transcription there in comparison, and it is really bad there.

Is it improved on next Premiere-Version, or is there a way to get “the transcription feature” outside of podcast.adobe.com - on which model/gpt is adobe’s podcast-feature working?

thanks!