NVIDIALaunchAI ModelsOctober 28, 2026

NVIDIA releases Nemotron 3.5 Transcribe open speech model

Read original source →huggingface.co

The open-weight 30B MoE (~4B active) handles 40-language transcription, translation for 31 languages, speaker attribution for up to 8 speakers, and 1-hour recordings in one checkpoint. It tops Hugging Face's Open ASR open-model leaderboard for English at 5.33% average WER, but works only on recorded audio, not live streams.

How this story unfolded

1 day · 5 reports · 4 community posts · 9 of 10 shown

  1. Oct 28
  2. Oct 29

More stories today

Open the live feed