Nvidia releases Nemotron 3.5 Transcribe for recorded speech

Nvidia on Oct. 28 released open weights for Nemotron 3.5 Transcribe, a model it says transcribes 40 languages, translates 31 to or from English, and labels up to eight speakers in recordings as long as one hour. The company says about 4 billion of its 30 billion parameters are active per token, and that the release is for recorded audio, not live streams. Quantized local versions are described as coming later. The same release includes Granary v2, whose public manifest Nvidia counts at 463,062 hours across 48 languages, separate from the model's full training mix.