How much does Nemotron 3.5 Transcribe cost?

It is free and open-weight, released by Nvidia on Wednesday alongside version 2 of its open source Granary speech dataset.

Can it transcribe live audio?

No. Transcribe works with recorded audio, not livestreams — a drawback Nvidia acknowledges even as it targets meeting and medical transcription apps.

What can it do besides transcribe 40 languages?

Nvidia says it can translate 31 languages to and from English, handle recordings up to an hour long, and tell up to eight speakers apart.

Audio & SpeechOpen-Source AITechCrunch AI1h agoOct 29, 2026, 01:00 JST

Nvidia releases Nemotron 3.5 Transcribe, free in 40 languages

Send on LINE

3 Key Points

What happened
Nvidia released Nemotron 3.5 Transcribe, a free open-weight model transcribing up to 40 languages, plus version 2 of its open source Granary speech dataset.
Why it matters
It targets developers building meeting and medical transcription apps, and Nvidia says the 30 billion parameter model runs on a single chip, from Jetson AGX Thor to Blackwell datacenter chips.

WHO IT HITSDevelopers and product teams building transcription features for business meetings and medical records get a free, customizable open-weight option — though anyone needing live, real-time speech must look elsewhere, since Transcribe handles only recorded audio.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, every morning.

Get these free every morning

Context & Analysis

Nvidia's release lands in a crowded stretch for AI voice. Mistral shipped Voxtral 3 earlier this month, ElevenLabs launched Realtime for instant speech in voice agents, and Salesforce released Launa-Omni, its voice model — a cluster of launches that shows how much weight vendors are putting on speech interfaces.

What distinguishes Nvidia's entry is its packaging rather than a single capability. Nvidia pairs the model with the updated Granary dataset and its NeMo Speech tools, telling developers they can tune apps to recognize an industry's jargon in each language. The 30 billion parameter model is positioned to run on a single chip across Nvidia's range, from the Jetson AGX Thor robotics controller and DGX Spark chip in new laptops to Blackwell datacenter chips, and it sits alongside smaller open-weight voice models like Parakeet at 1.1 billion parameters.

The gap Nvidia leaves open is real-time use: Transcribe handles recorded audio, not livestreams, so live voice-agent work appears to remain with rivals like ElevenLabs — at least for now.

FAQ

How much does Nemotron 3.5 Transcribe cost?: It is free and open-weight, released by Nvidia on Wednesday alongside version 2 of its open source Granary speech dataset.
Can it transcribe live audio?: No. Transcribe works with recorded audio, not livestreams — a drawback Nvidia acknowledges even as it targets meeting and medical transcription apps.
What can it do besides transcribe 40 languages?: Nvidia says it can translate 31 languages to and from English, handle recordings up to an hour long, and tell up to eight speakers apart.

TechCrunch AIRead Original Article

Send on LINE

Also reported by Hugging Face Blog, SiliconANGLE AI, THE DECODER

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Get it free

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.