NVIDIA Expands RTX Spark Software Ahead of Oct. 16 Launch

NVIDIA is rolling out software updates for RTX Spark ahead of the platform’s worldwide laptop launch on Oct. 16, expanding its focus from graphics to local artificial-intelligence agents, creative applications, developer tools and gaming.

RTX Spark combines a Blackwell RTX graphics processor with up to 6,144 CUDA cores, a Grace CPU with up to 20 cores, up to 128GB of unified memory and up to 1 petaflop of FP4 AI performance. Systems are planned from Acer, ASUS, Dell, HP, Lenovo, Microsoft, MSI and Gigabyte.

Updated agent applications include GitHub Copilot, Hermes, OpenClaw, Perplexity, Pi and WorkBuddy. The applications are designed to support local-model setup, hybrid routing between local and cloud models, and security features involving OpenShell and Microsoft Execution Containers. Microsoft has made Execution Containers generally available for running agents securely and persistently in the background.

“The personal computer is the ultimate tool, it’s my ultimate tool, and for a whole generation of people, it’s our ultimate tool,” NVIDIA founder and CEO Jensen Huang said. “What happens in the era of agents, when the agent is on your computer? It is now your personal assistant.”

NVIDIA Nemotron 3.5 Bolt 74B-A9B 24GB is scheduled for release Oct. 15. The mixture-of-experts model has 74 billion parameters, including 9 billion active parameters, and uses 2-bit quantization to fit within 24GB.

NVIDIA PAIR is moving out of beta with support for llama.cpp, which distributes AI workloads across compatible PCs on a local network. A Windows-native llama.cpp application will let users download and run models without the command line. NVIDIA says decision models supported by llama.cpp can make agent decisions up to 200 times faster than traditional agents.

NVIDIA and Adobe are reworking Photoshop and Premiere for RTX Spark, with up to 2x faster performance claimed across AI, editing, color and effects workflows. Adobe Chair and CEO Shantanu Narayen said the expanded partnership will make Adobe experiences “faster and more powerful than ever.”

NVIDIA Broadcast 2.3 will add Video Super Resolution, Video Frame Generation, offline file processing and Model Context Protocol controls. PyTorch, Python, CUDA and CUDA libraries are also ready for the platform, while Microsoft is making coding models such as MAI 1.1 Coding Flash available locally.

Hundreds of games are planned at launch, including Cyberpunk 2077, Fortnite, Genshin Impact, Hogwarts Legacy, Marvel Rivals, NBA 2K27 and World of Warcraft. A new DLSS 5 model is scheduled to debut in NBA 2K27 on Oct. 16. NVIDIA claims it is 2x faster and can increase frame rates by up to 1.4x.