The headline addition is Nemotron 3.5 Transcribe, an open weight speech model introduced today. One promptable model handles transcription, speaker labels and translation across 40 languages, including seven Indic languages, and developers can steer it with plain-language prompts and custom vocabulary. NVIDIA says it ranks No. 1 on the Hugging Face Open ASR leaderboard with about 4.3 billion active parameters, 80% fewer than the top open weight model, and the open weights let developers choose where audio is processed, locally or in the cloud.

For people running background AI agents on these machines, a new resource management feature in NVIDIA App lets users decide how GPU compute is shared between a foreground app and a background agent. It requires hardware-accelerated GPU scheduling to be enabled in Windows. Liquid AI's LFM3 open model family, spanning text, vision, audio and more, also gets day-zero support, with native support through llama.cpp and vLLM.

On the gaming side, S-GAME's Phantom Blade Zero launches tonight, Oct. 28, at 7 p.m. PT on Steam and the Epic Games Store with ray tracing and DLSS 4.5, including up to 6x frame generation. RTX Remix 1.6, released last week, brings path-traced rendering to Minecraft Java Edition 1.21.1 and adds native RTX Spark support, with the same upgrades arriving in Portal with RTX and the Half-Life 2 RTX demo.