NVIDIA RTX Spark Launches Oct 16 With DLSS 5 and a Huge App Lineup

NVIDIA's RTX Spark superchip lands October 16 with thousands of supported apps, DLSS 5, new local AI agents, and major Adobe performance gains.

Mark your calendar: October 16 is when NVIDIA's RTX Spark laptops officially hit shelves, and the ecosystem is already buzzing. NVIDIA and Microsoft announced the launch last week, and a wave of software updates is rolling out to make sure the hardware has plenty to do on day one.

If you haven't been following, RTX Spark is NVIDIA's bid to shake up the PC as we know it. It packs up to 128GB of unified memory and 1 Petaflop of AI compute into a single superchip, blending gaming PC, creative workstation, and personal AI supercomputer into one package. The full RTX and CUDA stack comes along for the ride, which means creators, developers, and players all get something out of it.

On the local AI front, things are moving fast. NVIDIA's Nemotron 3.5 Bolt 74B model, built specifically for agentic AI on RTX Spark, drops Thursday, October 15. NVIDIA PAIR is exiting beta with full release and llama.cpp support. Meanwhile, top agent apps like GitHub Copilot, Hermes, OpenClaw, Perplexity, Pi, and WorkBuddy are getting updates that simplify setup, boost security through OpenShell and Microsoft Execution Containers (MXC), and support hybrid local-cloud AI routing.

Creative apps are getting some love too. Adobe's flagship tools are in line for up to 2x performance gains, with NVIDIA and Adobe rebuilding GPU-accelerated Photoshop compositing and Premiere's playback, rendering, and effects from the ground up. NVIDIA Broadcast 2.3 adds video super resolution, video frame generation, offline file processing, and MCP controls for agents. ComfyUI now supports RTX Video frame generation and HDR to speed up AI video workflows, and DLSS Ray Reconstruction arrives in Blender 5.3 this November.

Developers aren't left out either. PyTorch, Python, CUDA, and the full CUDA library lineup already support RTX Spark, and coding agents like GitHub Copilot and Pi are optimized for it. llama.cpp gains decision model support plus a Windows app.

Gamers, here's your section. RTX Spark supports hundreds of popular games and anti-cheat systems. The DLSS 5 AI model, promising up to 2x performance, launches October 16, while the new DLSS Frame Generation Performance Mode is available now via the NVIDIA App beta starting October 13. The third-gen Nemotron 3.5 Bolt E4B 4GB small language model for NVIDIA ACE game characters arrives October 23, and RTX Remix gets DLSS 5 support the same day.

AI is changing how we use PCs, and local agents are having a moment. Projects like OpenClaw and Hermes are climbing GitHub's trending charts, and local LLMs have gotten smart enough to actually drive them. Downloads of local large language models have hit five times last year's total. NVIDIA is working across the local AI ecosystem to make RTX Spark the go-to platform for running agents, with a handy onboarding screen on first boot that lets you pick and install agent apps. Security comes via OpenShell and MXC sandboxing, and NVIDIA Switchyard, an open-source model routing library, helps balance requests between local and cloud models so you get near-cloud results without leaning on the cloud too hard. Hermes adds Switchyard later in October.

So, are you planning to pick up an RTX Spark machine, or are you waiting to see how the benchmarks shake out? Let us know in the comments.