NVIDIA RTX Spark Laptops Expand AI Capabilities with New Models and Tools

Local AI on the Go

RTX Spark laptops from ASUS, Dell, HP, Lenovo, Microsoft and MSI combine an NVIDIA Blackwell RTX GPU and Grace CPU in a single superchip, delivering up to 128 GB of unified memory and 1 petaflop of FP4 AI performance. This integration brings a personal AI supercomputer into a portable Windows device, enabling developers, creators and gamers to run demanding workloads locally.

New AI models arrive on the platform. NVIDIA Nemotron 3.5 Transcribe provides an open‑weight speech model that handles transcription, speaker labeling and translation in 40 languages with a single promptable interface. Liquid AI’s LFM3 family offers day‑zero support for text and audio tasks, while the model’s ~4.3 billion parameters keep memory usage low on edge hardware.

Agent tools benefit from tighter resource control. The updated NVIDIA App lets users balance GPU compute between foreground applications and background agents through the Hermes Agent, OpenShell and Switchyard integrations. This ensures that long‑running AI assistants do not starve critical workflows of graphics power.

Gaming gains a boost with Phantom Blade Zero launching with NVIDIA DLSS 4.5 and ray tracing, while Minecraft users can explore path‑tracing and RTX Remix 1.6. These features demonstrate how the same hardware can serve both creative and entertainment use cases.

Why it matters for GPU/AI infrastructure: The RTX Spark lineup proves that high‑performance AI inference can be delivered on thin‑and‑light form factors. For enterprises, this means lower latency, improved data privacy and the ability to deploy AI services at the edge without relying on centralized cloud resources.