NVIDIA is rolling out software updates for RTX Spark laptops ahead of their scheduled availability on Friday, October 16, giving buyers tools for local AI agents, editing and gaming. The platform combines up to 128GB of unified memory with 1 petaflop of AI compute and supports NVIDIA RTX and CUDA technologies. NVIDIA says thousands of games, apps and models will be available at launch.
Creative software readiness varies: AI optimizations for Photoshop and Premiere are available in beta, while Substance 3D improvements remain scheduled for later. NVIDIA says its work with Adobe will deliver performance gains of up to 2x across tasks including AI processing, editing and effects. Photoshop changes target compositing and image removal, while Premiere improvements cover video playback, rendering and features including Enhance Speech.
Read more: NVIDIA Adds Gears of War, E-Day to GeForce NOW With RTX 5080-Class Play
For local AI, NVIDIA Nemotron 3.5 Bolt 74B-A9B 24GB is scheduled for Thursday, October 15. The model has 74 billion parameters, activates 9 billion of them and uses 2-bit quantization to fit into 24GB. NVIDIA says users will be able to run it through tools including Ollama and LM Studio, or access it through agents such as OpenClaw and Hermes.
A setup screen on RTX Spark will let users select and install an agent app, with supported apps recommending a local model and configuring it for the machine. OpenClaw and GitHub Copilot already support Microsoft Execution Containers (MXC), while Hermes Agent is due to integrate MXC and granular OpenShell controls by the end of October. NVIDIA says its Switchyard library lets agents route requests between local and cloud models, and Hermes is scheduled to add it later this month.
NVIDIA PAIR leaves beta today, adding RTX Spark compatibility and llama.cpp support to a free, open source tool that distributes AI work among compatible networked PCs. NVIDIA Broadcast 2.3 is available now, with an Offline File Processor for applying effects to recorded material and controls that let local agents adjust settings and process files.
For gaming, NVIDIA says a DLSS 5 AI model with twice the speed arrives Friday, October 16. A new Performance Mode for DLSS Frame Generation becomes available today through the NVIDIA App beta. PyTorch, Python and CUDA are already available for RTX Spark developers.
RTX Spark laptop preorders opened October 7, with Microsoft's Surface Laptop Ultra among the machines using the platform. NVIDIA previously introduced PAIR in September as beta software for sharing inference work across computers on a local network.
