NVIDIA's RTX Spark laptops are here, and they want to run your AI tools and your games on the same chip

NVIDIA's pitch for its new RTX Spark laptops is simple: one thin machine for gaming, creative work and running AI locally. On October 28 the company said these laptops are now available from ASUS, Dell, HP, Lenovo, Microsoft and MSI, and it used the moment to show off a batch of AI updates alongside them.

What's inside

Each RTX Spark system pairs a Blackwell RTX GPU with a Grace CPU on a single superchip, with up to 128GB of shared memory. NVIDIA quotes up to 1 petaflop of FP4 AI performance, which is its way of saying these laptops are meant to run fairly large models without sending everything to the cloud.

The AI pieces

• Nemotron 3.5 Transcribe: a new open-weight speech model that handles transcription, speaker labels and translation in one go, across 40 languages. It's on Hugging Face now, with quantized versions for local use promised soon.
• LFM3 from Liquid AI: new open models with day-zero support on RTX Spark. The new 1.7B base model beats Liquid's earlier 2.6B one, and they run through llama.cpp and vLLM.
• GPU sharing controls: a new option in the NVIDIA App lets you decide whether your foreground app or a background AI agent gets priority on the GPU.

Why I care as a fan

That last feature is the interesting one for anyone who games. If you leave an AI assistant working in the background, you can now tell the laptop to keep your game smooth first. And a transcription model that runs on your own machine is handy for anyone who subtitles clips or keeps notes from streams and panels without uploading the audio anywhere.

NVIDIA also timed this with the PC launch of the action game Phantom Blade Zero, which supports ray tracing and DLSS 4.5 on Steam and the Epic Games Store. Whether these laptops are worth the price will come down to real-world reviews, but the direction is clear: local AI is being sold to players, not just developers.