Thin, Light and Built for Everything: NVIDIA RTX Spark Windows Laptops Gain New Models, Agent Tools and Games

RTX Spark laptops are on sale from ASUS, Dell, HP, Lenovo, Microsoft, and MSI, starting at $2,599. Each pairs a Blackwell RTX GPU and a Grace CPU on one superchip with up to 128GB of unified memory and up to 1 petaflop of FP4 AI performance, according to Nvidia.

Alongside, Nvidia introduced Nemotron 3.5 Transcribe, an open-weight speech model for transcription, speaker labels, and translation across 40 languages; announced day-zero support for Liquid AI's LFM3 open models; and added controls in the NVIDIA App to split GPU resources between foreground apps and background agents.

We're bullish on local AI, and this is a meaningful step. Running capable models on your own machine means sensitive audio, documents, and email don't have to leave the device. Nemotron 3.5 Transcribe is a good example: Nvidia says it handles transcription, speaker attribution, translation, and questions about the audio in one promptable model, and because the weights are open, developers decide where the audio is processed. For a clinic or a finance team, that choice matters.

The detail we find most telling is the GPU resource control. Nvidia is designing for a world where agents run in the background all day while you work in the foreground, and it's giving users a dial to decide who gets the compute. That's a quiet acknowledgment that personal agents are becoming a standing workload, not a demo.

The caveats. A $2,599 entry price keeps this in premium territory for now. Benchmark claims, like Transcribe ranking first on the Hugging Face Open ASR leaderboard, are Nvidia's own framing and worth watching as others test it. And the quantized versions that make Transcribe practical on these laptops are described as coming soon, not shipping today.

There's also a broader question about always-on local agents that triage inboxes and manage calendars, use cases Nvidia highlights for the LFM3 models. Local is better for privacy than cloud, but an agent with standing access to your mail is still an agent with standing access to your mail. The monitoring and permission questions don't disappear just because the compute is on your desk.

The update also spans gaming and creative work: Phantom Blade Zero launches with ray tracing and DLSS 4.5, and RTX Remix 1.6 brings path tracing to Minecraft Java Edition along with agent-assisted modding workflows. Nvidia also noted the Nemotron 3.5 Bolt E4B model, which runs multistep agent tasks in as little as 4GB of memory, and a 64GB DGX Spark now available from partners.