Post-Cutoff
Search For AI agents

NVIDIANemotron 3.5 familyCurrentSpeech and audioReleased 28 Oct 2026

NVIDIA Nemotron 3.5 Transcribe (30B-A4B)

Model idnvidia/NVIDIA-Nemotron-3.5-Transcribe-30B-A4B

PricePrice not published

Checked29 October 2026

Where it stands

No newer speech and audio model in our registry yet.

All speech and audio models

Before it from NVIDIA
NVIDIA NemotronLabs VoiceChat 11B (and PersonaPlex-7B)3 Aug 2026
Weights
Weights are public. Licence: OpenMDW-1.1. See the licence for commercial terms.

How to call it

5 ways to use NVIDIA Nemotron 3.5 Transcribe (30B-A4B), with the model id each provider expects.

Providers, model ids, endpoints and docs for NVIDIA Nemotron 3.5 Transcribe (30B-A4B)
ProviderModel idEndpointDocs
Hugging Face (open weights, BF16; FP8 and NVFP4 variants) nvidia/NVIDIA-Nemotron-3.5-Transcribe-30B-A4B huggingface.co
Hugging Face (optional DFlash2 draft model for speculative decoding) nvidia/NVIDIA-Nemotron-3.5-Transcribe-30B-A4B-DFlash2 huggingface.co
Self-hosted vLLM 0.32.0 (OpenAI-compatible server) with the NeMo Speech main branch (VLLM_PLUGINS=nemo_speechlm) huggingface.co
NVIDIA NIM / build.nvidia.com build.nvidia.com
NGC registry.ngc.nvidia.com

Pricing

Price not published.

No price source linked. Checked 29 October 2026.

Specs

Input
Audio and text
Output
Text
Context window
65,536 tokens
Max output
Not published
Open weights
Yes
Licence
OpenMDW-1.1
Released
28 October 2026
Status
Current
Type
Speech and audio

What stands out

What its maker marketed at launch, and what people found later, each with its source.

  • One model for transcription, diarization, translation and audio Q&A

    Speech-augmented LLM (ASR + speaker encoders feeding the hybrid Mamba-Transformer MoE Nemotron 3.5 Lightning backbone): verbatim or written-form transcripts, speaker tags, word timestamps (CTC head), translation to and from English, language ID, PII redaction and questions about the recording, chosen by the prompt; 40 languages; recordings up to 1 hour.

    Source: huggingface.co

  • Top open-weights scores on the Open ASR Leaderboard sets (vendor-run)

    NVIDIA’s runs (captured Oct 12, 2026): English Open ASR mean WER 4.71% (Qwen3-ASR-1.7B 4.94), long-form 10.98% (Parakeet-TDT-0.6B-v3 13.37), multilingual Open ASR (7 languages) 4.62% (Qwen3-ASR-1.7B 6.49), multi-speaker mean cpWER 15.80%. Independent leaderboard listing not checked.

    Source: huggingface.co

  • Contextual biasing by prompt

    Domain terms and names can be passed in the instruction (‘Context terms: ...’) to improve recognition of jargon.

    Source: huggingface.co

Notes

34B total parameters, ~4.3B active (3.5B language backbone + ~0.8B joint audio encoder). Offline (batch) model, not a streaming ASR service; for real-time use NVIDIA’s 0.6B streaming models (see nvidia-parakeet-canary). Runs on one H100 80 GB (BF16) and, quantized to NVFP4, on an RTX 5090; up to 2,238 s of audio per second on one RTX PRO 6000 Blackwell (NVFP4 + DFlash2, NVIDIA). Audio understanding and reasoning were evaluated only in English. Training: ~1.72M unique hours, including Granary v2 (public manifest 463,062 h, 48 languages) and data licensed from David AI.

Quick start (from the model card): vllm serve ./NVIDIA-Nemotron-3.5-Transcribe-30B-A4B with VLLM_PLUGINS=nemo_speechlm, then send the audio to the OpenAI-compatible endpoint with an instruction such as “Transcribe with punctuation and capitalization.” Keep the system prompt empty for transcription and translation; use temperature 0 and reasoning off. Technical report: https://research.nvidia.com/labs/nemotron/files/NVIDIA-Nemotron-3.5-Transcribe-Technical-Report.pdf