NVIDIA unveils Nemotron 3.5 Transcribe and Granary v2
On October 28, 2026, NVIDIA published a Hugging Face post introducing Nemotron 3.5 Transcribe and Granary v2. NVIDIA describes Transcribe as an open-weight, promptable speech model that combines transcription, speaker attribution, translation, and audio question answering in one checkpoint. The company says the model covers 40 transcription languages, translates 31 languages to or from English, attributes speakers in 13 languages, handles recordings up to an hour long, and labels up to eight speakers. NVIDIA also says Granary v2 is the open data-curation pipeline behind the model, with a public manifest covering 463,062 hours across 48 languages and 16 source corpora. All benchmark and performance results in the post are NVIDIA-reported claims rather than independently verified measurements.
Check the sources
This is a NextGen-AI briefing, not a reproduction of the discovery source. A newsletter may cover several developments; consult the evidence below for this briefing’s specific subject. The source date refers to the discovery entry, not necessarily the announcement date.