Alex Volkov(@altryne):This transcription model is absolutely insane. For some reason everyone is still using Whisper-3-large on OR, when this MAI-transcribe-2 is much more accurate. But the speed 😮 it deadass transcribed over 90 minutes of audio with diarization in 15 seconds! WTFFF.

Alex Volkov

@altryne

🎙️ Host of @thursdai_pod (get your AI news with us!) ✨ AI Evangelist with @wandb & @coreweave 🪄🐝 Opinions my own

Joined May 2007

2.6K Following 42.7K Followers

Alex Volkov@altryne

2026.09.03 22:36

This transcription model is absolutely insane. For some reason everyone is still using Whisper-3-large on OR, when this MAI-transcribe-2 is much more accurate. But the speed 😮 it deadass transcribed over 90 minutes of audio with diarization in 15 seconds! WTFFF.