NVIDIA RTX Spark Laptops Expand the Ecosystem: New AI Models, Agent Tools, and Gaming Support
Public- 28 Oct, 2026

NVIDIA RTX Spark laptops are now available from ASUS, Dell, HP, Lenovo, Microsoft, and MSI, starting at $2,599. These systems integrate an NVIDIA Blackwell RTX GPU and Grace CPU into a single superchip featuring up to 128GB of unified memory and 1 petaflop of FP4 AI performance. This architecture allows premium Windows laptops to handle desktop-class workloads, including complex creative projects and AI inference, while maintaining portability, with designs as thin as 14mm and as light as three pounds. The unified memory pool ensures that both creative applications and AI models can access resources efficiently, even when the device is unplugged, offering developers and creators on the go all-day battery life.
A major software update introduces NVIDIA Nemotron 3.5 Transcribe, an open-weight speech recognition model that consolidates transcription, speaker labeling, and translation into a single promptable system. It supports 40 languages, including seven Indic ones, and ranks highly on Hugging Face and Artificial Analysis benchmarks. With approximately 4.3 billion active parameters, it is 80% smaller than the top competing open model, making it highly practical for local hardware. Available in BF16, FP8, and NVFP4 formats, it runs on RTX Spark, DGX Spark, Jetson AGX Thor, and GeForce RTX 5090 GPUs. Developers can deploy it via NeMo-Speech.cpp or vLLM, allowing flexible processing from desktops to the cloud while correctly handling specialized vocabulary in healthcare and finance.
Liquid AI's LFM3 family offers compact, fast foundation models for text, vision, audio, and decision-making tasks. A new 1.7-billion-parameter base model outperforms its larger predecessor. These open-weight models support local workflows such as inbox triage, calendar management, and document reading through agents in Hermes, Pi, and OpenClaw. Native support via llama.cpp and vLLM allows developers to run and fine-tune these models directly on RTX Spark. Additionally, Nemotron 3.5 Bolt E4B 4GB enables natural conversations and multistep agent tasks with minimal memory use, powering local voice assistants and game characters on RTX Spark, DGX Spark, and Jetson devices, while the 74B-A9B variant handles more demanding agent workloads.

The gaming ecosystem also gets significant updates. S-GAME's Phantom Blade Zero launches on Steam and Epic Games Store with ray-traced effects and DLSS Dynamic Multi Frame Generation, delivering up to 6x frame boosts. RTX Remix 1.6 brings path tracing to Minecraft Java Edition 1.21.1, allowing creators to render realistic lighting and materials. The update also adds DLSS 5 with 3D-Guided Neural Rendering and native RTX Spark support to Portal with RTX and the Half-Life 2 RTX demo. These enhancements include a new Performance Mode for frame generation and options to speed up path tracing, ensuring high-fidelity visuals on mobile hardware.

RTX Remix 1.6 simplifies AI-powered modding with a redesigned ComfyUI workflow that generates physically based materials directly within the platform, eliminating the need for API setup. It includes batch processing for full projects and presets for less technical users, while NVIDIA agent skills allow compatible AI agents to assist in remastering projects. Furthermore, Black Forest Labs' FLUX 3 Image Klein, a 5.6-billion-parameter open-weight model, enables local image generation and editing with as little as 8GB of VRAM on RTX GPUs. Developers can fine-tune this model on RTX Spark using an NVIDIA NeMo AutoModel recipe to customize it for specific creative workflows, expanding the toolkit for local AI-driven content creation.