NVIDIA releases Nemotron 3.5 Transcribe open-weight speech model
NVIDIA introduced Nemotron 3.5 Transcribe, an open-weight, promptable speech model that combines transcription, speaker attribution, translation, contextual prompting, and audio understanding across 40 languages. Testingcatalog, citing NVIDIA AI, describes it as a 30B mixture-of-experts model with roughly 4B active parameters and reports that it ranked first in VoiceArena's 14-system diarization and ASR comparison. NVIDIA says the model is available on RTX Spark laptops.
The release consolidates several speech-processing functions into one locally runnable open-weight checkpoint, while its reported VoiceArena result provides a notable external comparison.
NVIDIA RTX Spark powers thin and light laptops built for everything that developers, creators and gamers do.
RTX Spark laptops are now available from ASUS, Dell, HP, Lenovo, Microsoft and MSI. These powerful systems combine an NVIDIA Blackwell RTX GPU and an NVIDIA Grace CPU in a single superchip with up to 128GB of unified memory and up to 1 petaflop of FP4 AI performance. That puts the NVIDIA CUDA platform, RTX graphics and the capabilities of a personal AI supercomputer inside premium Windows laptops built to go anywhere.
New and expected RTX Spark updates, coming soon, include:
- NVIDIA Nemotron 3.5 Transcribe, an open weight speech model for transcription, speaker attribution and translation is available on RTX Spark.
- The LFM3 family of open models with day-zero support.
- Resource management controls in NVIDIA App to allocate GPU resources between foreground apps and background agents.
- Phantom Blade Zero, S-GAME’s highly-anticipated action role-playing game, launches with ray tracing and NVIDIA DLSS 4.5.
- Experience Minecraft with path tracing and DLSS with RTX Remix1.6, available now for RTX Spark.
Turn Recorded Audio Into Useful Information With NVIDIA Nemotron 3.5 Transcribe
Introduced today, NVIDIA Nemotron 3.5 Transcribe is a new, open weight speech model that turns recorded conversations into useful information. Instead of needing multiple models for transcription, speaker labels and translation, developers can use one promptable model that handles all three, plus questions about the audio. It transcribes 40 languages, including seven Indic languages.