NVIDIA expands RTX Spark laptop lineup with agent tools, models and games

NVIDIA said RTX Spark laptops are now available from ASUS, Dell, HP, Lenovo, Microsoft and MSI, while the Windows platform is gaining new AI models, agent controls and game support. The update adds more evidence that the company is positioning RTX Spark as a single device category for local AI development, content creation and gaming.

Platform updates focus on local AI workflows

In a post published Oct. 28, 2026, NVIDIA said the RTX Spark 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. The company said the laptops are as thin as 14mm and as light as three pounds, and that 128GB configurations are on shelves starting at $2,599.

Among the new and upcoming software additions, NVIDIA said Nemotron 3.5 Transcribe is available on RTX Spark. The open-weight speech model is intended for transcription, speaker attribution and translation, and the company said it supports 40 languages, including seven Indic languages. NVIDIA also said the model is available on Hugging Face in BF16, FP8 and NVFP4 formats, with support for RTX Spark, DGX Spark, Jetson AGX Thor devices and GeForce RTX 5090 GPUs.

NVIDIA said the model ranks No. 1 on the Hugging Face Open ASR benchmark and No. 2 among open-weight models on Artificial Analysis, citing approximately 4.3 billion active parameters. Those benchmark claims were supplied by NVIDIA in the Oct. 28 post.

Agent controls and model support expand

NVIDIA said the latest NVIDIA App update adds resource management controls so users can allocate GPU resources between foreground applications and background agents. The company said the setting is intended to let users prioritize either a working app or a long-running agent, and that the feature requires hardware-accelerated GPU scheduling to be enabled in Windows Settings.

The company also highlighted Liquid AI's LFM3 family of open models, saying day-zero support is coming to RTX Spark. NVIDIA described LFM3 as a compact family spanning text, vision, audio, omni and decision models built on a shared backbone, with native support through llama.cpp and vLLM. NVIDIA said the models can be used for local workflows such as inbox triage, calendar management, document reading, speech transcription and translation, and support-ticket routing.

Gaming and modding support broaden

On the gaming side, NVIDIA said S-GAME's Phantom Blade Zero launches on Oct. 28 at 7 p.m. PT on Steam and the Epic Games Store with ray tracing and NVIDIA DLSS 4.5. NVIDIA said the title uses ray-traced lighting, reflections and shadows, and that DLSS Dynamic Multi Frame Generation can raise frame rates with up to 6x frame generation.

The company also said Minecraft support is available through RTX Remix 1.6, alongside updates for Portal with RTX and the Half-Life 2 RTX demo. NVIDIA said RTX Remix 1.6 adds native RTX Spark support, agent-assisted workflows, DLSS 5 with 3D-Guided Neural Rendering and up to 6x Multi Frame Generation. It also said the release includes a redesigned ComfyUI workflow for generating physically based materials directly inside RTX Remix.

What the update means for the platform

The Oct. 28 post builds on NVIDIA's earlier RTX Spark announcements from Oct. 16, when the company said the laptops were available from the same six manufacturers. NVIDIA's background materials describe RTX Spark as a Windows 11 platform that combines an Arm-based NVIDIA Grace CPU and integrated Blackwell RTX GPU for local AI development, content creation and gaming. The company separately distinguishes RTX Spark from DGX Spark, which is aimed at AI development and uses DGX OS.

As presented by NVIDIA, the latest update extends the platform beyond hardware availability into software and model support, with changes aimed at developers, creators and gamers using the same local machine for AI workloads and interactive applications.