RTX Spark Goes On Sale Today: What Nvidia Confirms

Nvidia’s RTX Spark is officially on sale today, October 16, 2026. The “superchip” that Nvidia calls RTX Spark — officially named RTX Spark N1X on its own spec sheet — is now shipping in slim laptops and compact desktops from six confirmed launch partners, Nvidia said in a GeForce news post and on its RTX Spark product page. The pitch: a Blackwell RTX GPU and a Grace CPU sharing up to 128GB of unified memory, built to run large AI models locally on a Windows 11 PC that still plays games. Here is what Nvidia itself has confirmed about the hardware, who is selling it, and — just as important — what Nvidia has not yet said about price.

RTX Spark Launch Day: What’s Actually Shipping

Nvidia’s own GeForce news post announcing the launch states plainly that “availability begins on October 16th from our OEM laptop partners, and preorders are available now.” That makes today the first day RTX Spark hardware is buyable at retail rather than something on a keynote slide. Nvidia’s RTX Spark product page backs that up, listing live product cards for six partner laptops alongside an “Email Me When Available” waitlist for anyone whose configuration or region isn’t stocked yet.

RTX Spark is built around what Nvidia calls a “superchip” — a single package pairing an Nvidia Blackwell RTX GPU with an Nvidia Grace CPU over Nvidia’s NVLink-C2C chip-to-chip interconnect, all running Windows 11. On its dedicated specifications page, Nvidia names this exact combination the RTX Spark N1X, which lines up with the “N1X” naming that’s been circulating in searches ahead of launch. Nvidia frames the whole platform as one device for three audiences at once: creators pushing large 3D scenes, developers running AI agents locally, and gamers who still want ray tracing and DLSS on the same machine.

What Is RTX Spark, Exactly?

Think of RTX Spark as Nvidia trying to put a miniature AI supercomputer inside a normal-looking Windows laptop. Instead of a conventional discrete-GPU-plus-separate-system-memory layout, the RTX Spark N1X superchip shares a single pool of unified memory between its Blackwell GPU and Grace CPU, the same architectural idea Nvidia uses in its DGX Spark desk units, just tuned for thin-and-light laptops and small desktops rather than developer workstations. Nvidia’s product page says the design is meant to let a single PC “build worlds, not workarounds,” handling 3D scenes up to 90GB in size while keeping a ray-traced viewport responsive — and separately, to run AI models up to 120 billion parameters with a 1 million token context window locally, no cloud subscription required.

That dual identity is the whole premise of RTX Spark: it is pitched as a gaming laptop that happens to also be a local AI workstation, rather than a workstation that happens to also game. Nvidia’s messaging leans hard on “on-device” AI — your data, in Nvidia’s words, “stays on your device, with the cloud available when you need it” — while the gaming side gets the full current RTX stack on day one: ray tracing, DLSS 5, Reflex, G-SYNC, and RTX Remix, according to the GeForce news post.

RTX Spark N1X Specs: What Nvidia Confirms

Nvidia’s specifications page breaks the N1X superchip into three configurations — two laptop tiers and one desktop tier — rather than a single fixed spec. All three share the same 256-bit unified memory interface and Windows 11 support, but core counts, power envelope, and AI throughput scale up or down depending on the tier:

All three configurations get fourth-generation RT cores, fifth-generation Tensor Cores, a ninth-generation NVIDIA encoder, a sixth-generation decoder, PCI Express Gen 5, and support for NVIDIA RTX, CUDA, and the full NVIDIA AI software stack, per Nvidia’s full spec sheet. The laptop tiers also list a dedicated NVIDIA NPU, which the desktop configuration’s listing does not show. For context on what that core-count scaling actually buys in rendering terms, it sits well below Nvidia’s top discrete desktop GPUs — see our RTX 5090 vs. RTX 4090 comparison for a sense of where flagship desktop Blackwell silicon lands by contrast. RTX Spark’s GPU is not trying to compete there; it is trading raw core count for a shared, much larger unified memory pool that a discrete gaming GPU with dedicated GDDR7 — the kind we cover in our GDDR7 explainer — simply cannot offer at these capacities.

Who’s Selling RTX Spark Today

Nvidia’s RTX Spark product page lists six launch-partner laptops available today, each linking out to a manufacturer-filtered listing on NVIDIA’s own marketplace:

Nvidia’s GeForce news post refers more generally to “our OEM laptop partners” without repeating the full list in that post, but the product page’s manufacturer filters on the NVIDIA Marketplace RTX Spark listing confirm all six by name. Nvidia has not stated whether additional OEMs or desktop-form-factor RTX Spark systems will follow; the desktop tier that appears in the spec sheet is not currently attached to a named desktop product on the retail page.

What RTX Spark Costs — And What Nvidia Hasn’t Said

This is the part buyers should slow down on: as of publication, Nvidia has not published an official price for RTX Spark, for any configuration, on either its GeForce news post or its RTX Spark product page. The product page’s own call to action for several configurations is “Email Me When Available,” not a buy button with a listed price, which signals Nvidia is routing buyers to its OEM partners’ own storefronts for final retail pricing rather than setting one official number itself.

Third-party outlets and retailers have published their own price estimates for individual RTX Spark laptops ahead of and around launch day, with figures that have circulated in the general ballpark of $2,000–$3,000 depending on configuration and memory tier. Those numbers are not confirmed by Nvidia and should be treated as third-party estimates only — this article does not repeat specific third-party price figures as fact, because Nvidia itself has not stated them. Anyone comparing prices today should check the manufacturer-specific links Nvidia itself provides on its marketplace listing, since actual retail pricing will come from ASUS, Dell, HP, Lenovo, Microsoft, and MSI directly, and may vary by region, memory configuration, and retailer.

DLSS 5, Local AI, and What Runs on Day One

Nvidia is leaning on software, not just silicon, to sell RTX Spark on launch day. The GeForce news post says RTX Spark supports the complete current suite of DLSS 5 technologies — Super Resolution, Ray Reconstruction, Dynamic Multi Frame Generation, and Nvidia’s newest addition, 3D-Guided Neural Rendering — alongside Reflex, G-SYNC, ray tracing, and RTX Remix from day one. Nvidia says “hundreds” of top games are playable on RTX Spark starting today, including titles from 2K Games, Activision, Blizzard, CD PROJEKT RED, Electronic Arts, Embark Studios, Epic Games, id Software, KRAFTON, Microsoft, HoYoverse, NetEase, Riot Games, Tencent, Wargaming, and Warner Bros. Games, per Nvidia’s own list. Several major anti-cheat systems — including BattleEye, EA Javelin Anticheat, Easy Anti-Cheat, RICOCHET Anti-Cheat, Vanguard, and Tencent Anti-Cheat Expert — are also supported on RTX Spark according to Nvidia, which matters for players who rely on competitive online titles. For a closer look at what the latest DLSS generation changes in practice, see our coverage of DLSS 4.5 on GeForce NOW, since several of the same controls and technologies carry forward into DLSS 5 on RTX Spark.

On the AI side, Nvidia’s pitch is that RTX Spark’s unified memory lets a single laptop run AI models that would otherwise need a dedicated workstation or cloud subscription. Nvidia’s product page specifically claims RTX Spark can run models “up to 120B parameters with 1M context,” built on CUDA, TensorRT, and NVFP4 — “the same stack the world’s AI runs on,” in Nvidia’s words — so that models built locally on RTX Spark are meant to run unchanged on other Nvidia GPUs. That’s a notably different pitch from a typical gaming laptop spec sheet, and it is the core reason Nvidia is positioning RTX Spark as a new category rather than simply another RTX-branded laptop refresh.

What to Know Before You Buy Today

If you’re looking at RTX Spark on its launch day, a few practical points worth keeping in mind, based strictly on what Nvidia has published:

• Configurations vary more than usual. The jump from the base laptop tier (5,120-core GPU, 18-core CPU, up to 32GB of memory, 827 teraflops) to the top laptop tier (6,144-core GPU, 20-core CPU, up to 128GB, 1 petaflop) is large. If local AI model size is your priority, the memory tier matters as much as the GPU core count.
• Pricing is not yet standardized. With no official Nvidia price sheet, expect the same configuration to carry different prices across ASUS, Dell, HP, Lenovo, Microsoft, and MSI, and across regions.
• Game support is broad but staggered. Many titles are playable on RTX Spark starting today, but Nvidia’s own list notes some games — including Call of Duty, PUBG: BATTLEGROUNDS, and Arena Breakout: Infinite — are “coming soon” or landing on RTX Spark later in 2027, not on day one.
• This is a new platform, not a GeForce RTX 50-series refresh. RTX Spark’s unified-memory superchip design is architecturally distinct from a conventional discrete GeForce RTX laptop GPU, so direct frame-rate comparisons to existing RTX 40- or 50-series gaming laptops aren’t something Nvidia has published either.

What’s Next for RTX Spark

Nvidia has already flagged several updates arriving in the weeks after launch rather than on day one. DLSS 5 with 3D-Guided Neural Rendering is scheduled to land in Hogwarts Legacy on November 5, and in Where Winds Meet in late November, both running on RTX Spark as well as GeForce RTX desktops and laptops. World of Warcraft: Forever is due to add support the day it launches, November 4. EA Javelin Anticheat is expected on RTX Spark devices in early 2027, and Nvidia says both Call of Duty and PUBG: BATTLEGROUNDS are planned for the platform in 2027. None of this is speculative — it comes directly from Nvidia’s own GeForce news post — but it does mean the RTX Spark software picture on October 16 is a starting point, not the finished lineup.

The bigger open question is pricing. Until Nvidia or its OEM partners publish official numbers, anyone shopping RTX Spark today will be comparing retailer listings rather than a single manufacturer price sheet. Expect that to firm up over the coming days as ASUS, Dell, HP, Lenovo, Microsoft, and MSI list final configurations on their own storefronts, linked from Nvidia’s own RTX Spark configuration guide.

Written byMara Lindqvist

Mara Lindqvist edits the hardware desk. She covers graphics cards, processors, memory and storage, the foundries and chip designers behind them, and what the numbers on a spec sheet mean for people choosing a PC. Specifications in her stories come from manufacturer spec pages and datasheets.

More from Mara LindqvistAll Hardware coverage