VOICE & AUDIO · WORTH A LOOK
NVIDIA Nemotron 3.5 Transcribe
NVIDIA's open-weight speech model (34B total / 4.3B active parameters): transcription in 40 languages, translation to and from English,…
What NVIDIA Nemotron 3.5 Transcribe does
NVIDIA's open-weight speech model (34B total / 4.3B active parameters): transcription in 40 languages, translation to and from English, speaker-attributed transcripts, word timestamps, contextual vocabulary and audio question answering, for recorded audio up to one hour (offline, not live). Released 28 Oct 2026 on Hugging Face, NGC and build.nvidia.com; served via NeMo Speech with vLLM, SGLang-Omni or Transformers.
Good for
- Private, self-hosted meeting and call transcription with speaker labels
- Transcripts with your own jargon and names supplied as context
Try this in 10 minutes
- Private, self-hosted meeting and call transcription with speaker labels. Use the free version: pick one real example from your own business and try just that.
- Transcripts with your own jargon and names supplied as context. Put the result next to how you do it today. Is it faster, cheaper or better?
- Sign up (free plan or trial if available) and try it on one real task from this week.
Mini-tasks based on common small-business uses for this tool.
Honest take
Some claims not independently checked
- Model card: huggingface.co (website field set to the NeMo Speech repo because add_tool.py would merge a huggingface.co URL into the Hugging Face entry).
- NVIDIA's demo video (a captioned film clip) showed visible speaker-label and word slips: video-notes/2026-10-29_nvidia-nemotron-3-5-transcribe-demo.md.
- Posted by @NVIDIAAI.
What we couldn’t verify
- Leaderboard claims are NVIDIA's.
Pricing
Free open weights (OpenMDW-1.1 licence, commercial use allowed); you pay for your own GPU (reference setup one H100 80 GB) or a hosted provider.
Prices as listed by the maker when we last checked (2026-10-29); usually in US dollars. Check the website for current prices.
Latest update
- 2026-10-28: Released (open weights). Model card: huggingface.co
Is NVIDIA Nemotron 3.5 Transcribe worth it? Quick answers
Is NVIDIA Nemotron 3.5 Transcribe worth it?
NVIDIA Nemotron 3.5 Transcribe is worth a look if it matches a job you already do (Voice & audio). It is free or open source, so the main cost is your time to set it up. Some of the maker's claims have not been independently checked.
How much does NVIDIA Nemotron 3.5 Transcribe cost?
Free open weights (OpenMDW-1.1 licence, commercial use allowed); you pay for your own GPU (reference setup one H100 80 GB) or a hosted provider.
What is NVIDIA Nemotron 3.5 Transcribe used for?
NVIDIA's open-weight speech model (34B total / 4.3B active parameters): transcription in 40 languages, translation to and from English,… Typical uses: private, self-hosted meeting and call transcription with speaker labels; transcripts with your own jargon and names supplied as context.