MetaPatchAI home
AI

NVIDIA's RTX Spark Laptops Get an Agent-Era Update: GPU Sharing for AI Agents, LFM3 Models and Path-Traced Minecraft

Source: NVIDIA Blog

Compiled by MetaPatchAI · Published , 7:13 AM ET

NVIDIA's RTX Spark Laptops Get an Agent-Era Update: GPU Sharing for AI Agents, LFM3 Models and Path-Traced Minecraft

Oct 30, 2026: What happens when the AI agent running in the background of your laptop wants the same GPU as the game in the foreground? NVIDIA's answer, as of this week, is a slider. In an Oct. 28 post on the NVIDIA Blog, the company detailed a batch of updates for its RTX Spark Windows laptops: new resource-management controls that split GPU power between apps and background agents, day-zero support for Liquid AI's LFM3 open models, new small Nemotron models, and RTX Remix 1.6, which brings path tracing to Minecraft Java Edition.

Quick recap: what RTX Spark is

According to NVIDIA, RTX Spark laptops "combine an NVIDIA Blackwell RTX GPU and an NVIDIA Grace CPU in a single superchip with up to 128GB of unified memory and up to 1 petaflop of FP4 AI performance." NVIDIA says the laptops are available now from ASUS, Dell, HP, Lenovo, Microsoft and MSI, "starting at $2,599, with 128GB configurations on shelves," and are "as thin as 14mm and as light as three pounds."

The pitch is a single pool of memory shared by the CPU, GPU and AI models, so work that "used to call for a desktop workstation" can run on a laptop, even unplugged.

The new part: deciding who gets the GPU

The most practical update is a resource management feature in the NVIDIA App. NVIDIA says users who have "long-running agents working alongside their favorite applications" can now choose to "prioritize their foreground app for the best performance, or give their background-running agent more resources to get work done faster."

How to find it, per NVIDIA: System > Advanced in the NVIDIA App. NVIDIA says the feature "requires hardware-accelerated GPU scheduling to be enabled in Windows Settings."

The control is small, but it shows where PCs are heading: NVIDIA is treating "an agent working in the background while you game or edit" as a normal workload to plan for.

New local models

  • Liquid AI LFM3: NVIDIA says the LFM3 family "spans text, vision, audio, omni and decision models built on one shared backbone," with open weights. According to NVIDIA, Liquid AI's "new 1.7-billion-parameter base model outperforms its previous 2.6-billion-parameter base model." That's a vendor claim we haven't independently tested. NVIDIA lists support through llama.cpp and vLLM, plus local fine-tuning.
  • Nemotron 3.5 Bolt E4B 4GB: NVIDIA says this one enables "natural conversations and multistep agent tasks with as little as 4GB of memory" and can power "local voice assistants, game characters and embedded interfaces." The game-character angle is the one to watch for gaming. A larger Nemotron 3.5 Bolt 74B-A9B 24GB targets heavier local agent workloads.
  • FLUX 3 Image Klein: Black Forest Labs' open-weights image model, which NVIDIA says runs in optimized versions "with as little as 8GB of VRAM."
  • Nemotron 3.5 Transcribe: NVIDIA's new open speech model also features in the post. We covered it separately.

For gamers: RTX Remix 1.6 and path-traced Minecraft

NVIDIA says RTX Remix 1.6, released the week before the post, lets creators and players "render Minecraft Java Edition 1.21.1 through RTX Remix with DLSS-accelerated path tracing." The update also adds what NVIDIA calls "DLSS 5 with 3D-Guided Neural Rendering, up to 6x Multi Frame Generation, native RTX Spark support and agent-assisted workflows."

According to the post, those upgrades come to Portal with RTX and the Half-Life 2 RTX demo starting Oct. 28. For modders, a redesigned ComfyUI workflow can now generate physically based materials inside RTX Remix "with no application programming interface setup," and "compatible AI agents can work directly with RTX Remix" through NVIDIA agent skills.

The post also notes that Phantom Blade Zero launched with ray tracing and DLSS 4.5. We covered that launch separately.

Also mentioned

  • DGX Spark 64GB is now available to buy from manufacturer partners, according to NVIDIA.
  • NVIDIA describes a GTC Berlin demo in which agents running on RTX Spark helped train Hugging Face's Microduck robot, using Hermes Agent, Nemotron 3.5 Bolt and Isaac Lab/Isaac Sim.

What's confirmed / not confirmed

Confirmed (NVIDIA Blog, Oct. 28, 2026): - RTX Spark laptops are available from ASUS, Dell, HP, Lenovo, Microsoft and MSI, starting at $2,599 (NVIDIA's figure) - The NVIDIA App's resource management control is under System > Advanced and requires hardware-accelerated GPU scheduling - Day-zero LFM3 support; Nemotron 3.5 Bolt E4B 4GB and 74B-A9B 24GB; FLUX 3 Image Klein optimized versions - RTX Remix 1.6 supports Minecraft Java Edition 1.21.1; Portal with RTX and the Half-Life 2 RTX demo get the new upgrades

Not confirmed / vendor claims only: - Performance claims (1 petaflop FP4, LFM3 1.7B beating 2.6B, "all-day battery life") are NVIDIA's and Liquid AI's and haven't been independently tested - NVIDIA's own list is labeled "new and expected RTX Spark updates, coming soon," so some items may not be live on every machine yet - Pricing for specific laptop models and regional availability

Sources

Related

Up: AI