Nvidia unveils RTX Spark software push for AI & games

Nvidia has outlined a broad set of software updates for its RTX Spark laptops ahead of their retail launch, spanning local AI agents, creative applications, developer tools and games.

The hardware will ship with support for a range of AI, content creation and coding software designed to run locally on the device, alongside updates from partners including Microsoft, Adobe and GitHub.

One of the biggest changes is a push to make local AI agents easier to install and use. On first boot, users will see a getting-started screen that lets them choose and install an agent app, while OpenShell and Microsoft Execution Containers add sandboxing and operating system-level security.

Applications including GitHub Copilot, Hermes Agent, OpenClaw, Perplexity, Pi and WorkBuddy are being updated to recommend local models and settings for RTX Spark systems and automate more of the setup process.

Nvidia is also moving its Personal AI Router, or PAIR, out of beta. The software links compatible PCs on a local network and distributes AI workloads across those machines. The latest version adds support for llama.cpp, direct model downloads from Hugging Face and custom inference settings.

Nvidia is also releasing a new large language model, Nvidia Nemotron 3.5 Bolt 74B-A9B 24GB, for use with local agents on RTX Spark. The model is built to run locally and will be available through tools including llama.cpp, vLLM, Ollama, Unsloth and LM Studio, as well as through agents such as Hermes and OpenClaw.

Creative software

Adobe is preparing optimised versions of Photoshop, Premiere and Substance 3D for RTX Spark. The changes are expected to bring as much as a twofold performance increase in some AI, editing, colour and effects tasks. Beta builds for Photoshop and Premiere are already available, while Substance 3D updates are due later.

The work includes GPU-accelerated compositing in Photoshop, and faster playback, rendering and colour grading in Premiere. Substance 3D Painter is also being updated to speed up texture creation, including baking mesh maps and handling large projects.

Nvidia Broadcast 2.3 is also being released with native support for RTX Spark. The update adds Video Super Resolution, Video Frame Generation, offline file processing and controls that let local AI agents adjust settings and process files through the app.

ComfyUI is adding RTX Video Frame Generation and HDR support, along with EXR export. Nvidia says the combination is intended to help creators handle AI-generated video within a single workflow.

Blender has released a beta version of Blender 5.3 with compatibility for RTX Spark, with DLSS Ray Reconstruction support due later in that version. Nvidia says the addition is intended to improve viewport interactivity for artists working in the 3D editor.

Developer stack

Nvidia and Microsoft have also focused on getting developer tools ready for the platform. Top frameworks, including PyTorch, Python, CUDA and CUDA libraries, are available for RTX Spark, with new support on Windows on Arm.

The rollout includes what Nvidia describes as the first PyTorch-CUDA build for Windows on Arm, the first Python build native by default on Windows on Arm with Python 3.15, and the first native Conda installer for the platform with more than 2,000 packages. Nvidia says 97% of the top 15,000 Python wheels are ready.

GitHub Copilot is being updated with support for Microsoft Execution Containers and local model routing, while Microsoft is making coding models available locally and adding llama.cpp as a backend to Windows ML. Llama.cpp is also getting support for decision models and a new Windows-native companion application called Llama Windows.

Nvidia says decision models allow AI agents to decide what action to take next much faster than traditional agent systems, including in tasks such as classification and action selection.

Games support

For gaming, RTX Spark laptops will launch with support for hundreds of titles, including major multiplayer and single-player games, as well as anti-cheat support. Nvidia is also bringing its wider graphics software stack, including DLSS, Reflex, Remix, G-SYNC and ray tracing, to the platform.

DLSS 5, with what Nvidia calls 3D-Guided Neural Rendering, will arrive at launch, and a faster DLSS 5 AI model is being introduced first in NBA 2K27. Nvidia says the model will later be used in Hogwarts Legacy and Where Winds Meet.

Modding tools are also being updated. RTX Remix 1.6 will introduce a revised texture generation workflow through ComfyUI Version 2, moving AI tools into a dedicated sidebar and adding batch processing and presets for modders and artists.

Nvidia is also releasing Nemotron 3.5 Bolt E4B 4GB, a smaller language model for AI game characters and voice assistants. The model is designed for local use with a low memory footprint, while leaving room for applications and other models on the same system.

The updates show Nvidia positioning RTX Spark as a Windows laptop platform that can handle gaming, creative software and AI work on one machine, rather than shifting more of that workload to cloud services.