NVIDIA releases Nemotron 3.5 Transcribe as an open-weight model
One 30B MoE checkpoint handles transcription in 40 languages, speaker attribution and translation.
NVIDIA introduced Nemotron 3.5 Transcribe, an open-weight, promptable 30B mixture-of-experts model with about 4B active parameters.
It brings transcription in 40 languages, speaker attribution, translation, contextual prompting and audio understanding into a single checkpoint. NVIDIA showed it keeping track of who said what while several people talk at once.
Why it mattersAn open model that handles crosstalk and many languages in one package gives developers a strong alternative to closed speech APIs.
