Nvidia on Wednesday released Nemotron 3.5 Transcribe, a free, open-weight model that transcribes audio in up to 40 languages. The model, with 30 billion parameters, is designed for developers building applications for business meetings and medical transcription, handling recordings up to an hour long and distinguishing up to eight speakers. It can translate 31 languages to and from English and is claimed to outperform rivals on benchmarks. A key limitation is that Transcribe works with recorded audio, not live streams.