NVIDIA AI Updates: October 29, 2026
1. NVIDIA Releases Nemotron 3.5 Transcribe, One Open Model for ASR, Diarization, Translation, and Audio Q&A
NVIDIA released Nemotron 3.5 Transcribe, an open-weight promptable speech model on Hugging Face and ModelScope that combines transcription (40 languages), translation (31 languages to or from English), speaker attribution (13 languages, up to eight speakers), and audio question answering in one checkpoint, for recordings up to an hour long. It pairs a speech encoder with a hybrid Mamba-Transformer MoE language model, activating about 4B of 30B parameters per token. NVIDIA reports first place among open-weight models on the Hugging Face Open ASR English leaderboard (5.33% average WER), a tie for first among open-weight models on Artificial Analysis (2.8% AA-WER), and the lowest diarization error rate on David AI’s 21-language leaderboard. The checkpoint serves with vLLM today, with NVFP4 quantized versions and NeMo-Speech.cpp local deployment coming soon. NVIDIA also released the Granary v2 multilingual dataset and curation pipeline plus NeMo Speech fine-tuning recipes; in a medical case study, fine-tuning cut WER on held-out speakers by 45.8% relative to the best prompt. The model targets recorded audio, not live streaming. Source
2. RTX Spark Laptops Ship From Six OEMs, NVIDIA App Adds GPU Sharing Controls for Background Agents
NVIDIA said RTX Spark Windows laptops, which pair a Blackwell RTX GPU and Grace CPU on one superchip with up to 128GB of unified memory and up to 1 petaflop of FP4 performance, are now available from ASUS, Dell, HP, Lenovo, Microsoft, and MSI starting at $2,599, with 128GB configurations on shelves. A new resource management feature in the NVIDIA App (System > Advanced) lets users decide whether foreground apps or long-running background agents get priority on shared GPU compute; it requires hardware-accelerated GPU scheduling in Windows. RTX Remix 1.6 also adds NVIDIA agent skills so compatible AI agents can drive remastering projects, along with a redesigned ComfyUI workflow for generating materials. Source