{"schema":"postcutoff/model@1","as_of":"2026-11-13T13:49:00+01:00","url":"https://postcutoff.com/m/nemotron-3-5-transcribe/","md":"https://postcutoff.com/m/nemotron-3-5-transcribe/index.md","disclosure":{"written_by":"AI agents (Claude Opus 5.5 in Claude Code)","editor":"human editor","policy":"https://postcutoff.com/about/"},"license":null,"id":"nemotron-3-5-transcribe","name":"NVIDIA Nemotron 3.5 Transcribe (30B-A4B)","org":"NVIDIA","family":"Nemotron 3.5","released":"2026-10-28","status":"current","type":"audio/speech","modality_in":["audio","text"],"modality_out":["text"],"open_weights":true,"model_license":"openmdw-1.1","context_window":65536,"max_output":null,"knowledge_cutoff":null,"pricing":null,"price_line":"Price not published","access":[{"provider":"Hugging Face (open weights, BF16; FP8 and NVFP4 variants)","model_id":"nvidia/NVIDIA-Nemotron-3.5-Transcribe-30B-A4B","url":"https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Transcribe-30B-A4B"},{"provider":"Hugging Face (optional DFlash2 draft model for speculative decoding)","model_id":"nvidia/NVIDIA-Nemotron-3.5-Transcribe-30B-A4B-DFlash2","url":"https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Transcribe-30B-A4B-DFlash2"},{"provider":"Self-hosted vLLM 0.32.0 (OpenAI-compatible server) with the NeMo Speech main branch (VLLM_PLUGINS=nemo_speechlm)","docs":"https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Transcribe-30B-A4B"},{"provider":"NVIDIA NIM / build.nvidia.com","url":"https://build.nvidia.com"},{"provider":"NGC","url":"https://registry.ngc.nvidia.com/orgs/0767305323357365/teams/transcribe-3_5-ea/models/nvidia-nemotron-3-5-transcribe-30b-a4b"}],"capabilities":[{"name":"One model for transcription, diarization, translation and audio Q&A","detail":"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.","first":false,"discovered":"launch","source":"https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Transcribe-30B-A4B"},{"name":"Top open-weights scores on the Open ASR Leaderboard sets (vendor-run)","detail":"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.","first":false,"discovered":"launch","source":"https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Transcribe-30B-A4B"},{"name":"Contextual biasing by prompt","detail":"Domain terms and names can be passed in the instruction ('Context terms: ...') to improve recognition of jargon.","first":false,"discovered":"launch","source":"https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Transcribe-30B-A4B"}],"entry":"2026-10-28-nvidia-nemotron-3-5-transcribe","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.","verified":"2026-10-29","body_md":"Quick start (from the model card): `vllm serve ./NVIDIA-Nemotron-3.5-Transcribe-30B-A4B` with `VLLM_PLUGINS=nemo_speechlm`, then send the\naudio to the OpenAI-compatible endpoint with an instruction such as \"Transcribe with punctuation and capitalization.\" Keep the system\nprompt empty for transcription and translation; use temperature 0 and reasoning off. Technical report:\nhttps://research.nvidia.com/labs/nemotron/files/NVIDIA-Nemotron-3.5-Transcribe-Technical-Report.pdf","page_url":"https://postcutoff.com/m/nemotron-3-5-transcribe/","standing":{"kind":"speech and audio","newer_count":0,"newer":[],"before":{"id":"nemotron-voicechat","name":"NVIDIA NemotronLabs VoiceChat 11B (and PersonaPlex-7B)","org":"NVIDIA","released":"2026-08-03","page_url":"https://postcutoff.com/m/nemotron-voicechat/"},"after":null,"weights":"public","licence":"openmdw-1.1","licence_permissive":false,"summary":"Weights are public. Licence: OpenMDW-1.1. See the licence for commercial terms."}}