On October 7, 2026, ASUS opened pre-orders for a trio of machines, the ProArt P16 and P14 laptops and the GR1X mini PC, all built on NVIDIA's new RTX Spark platform, with a promise that serious AI work can now live on a desk or in a bag rather than in a distant data center. The pitch rests on memory, up to 128GB of it, and on the old human preference for owning one's tools instead of renting them by the use. The figures come from the makers themselves and await the verdict of independent testing, a reminder that every new machine arrives first as a promise. Shipping is expected from October 16
ASUS Opens Pre-Orders for NVIDIA RTX Spark ProArt PCs Built for Local AI Agents
Intelligence you can truly own, rather than simply rent
Mark
So what is actually new here? Laptops that run AI aren't unusual.
Mimi
The memory. ASUS says up to 128GB of unified memory, which it argues lets bigger models run locally that wouldn't fit on a conventional PC.
Mark
And the appeal is avoiding cloud charges?
Mimi
Yes, no per-token or per-generation fees, and prompts and files staying on the device. The agents from Nous Research and ComfyUI are built around that local idea.
Luke
I'd flag that the 1 petaflop and 128GB figures are manufacturer claims. This is a press release, and there are no independent benchmarks in it.
Mark
Does the 128GB mean any large model will run?
Luke
No. ASUS itself says it depends on architecture, quantization, context length and available memory. That's a real qualifier.
Mimi
The quotes from Rolnick and Yan are also from partners with a stake in the products, so they read as endorsements.
Mark
What about price?
Mimi
Canadian pricing starts at C$3,499 for the P14 and C$4,999 for the P16, shipping around October 16. The 48GB and 64GB models come later.
Luke
And the release doesn't say what the 128GB configurations cost, or give pricing outside Canada. That's still unknown.
El Pulso
- Pre-orders for ASUS's ProArt P16, P14 and GR1X are open, with shipping estimated from October 16, 2026.
- Each machine promises up to 1 petaflop of AI performance and up to 128GB of unified memory, claims no independent test has yet confirmed.
- The pitch targets rising cloud bills, pitching local generation without per-token charges and data that never leaves the device.
- Agent software from Nous Research and ComfyUI aims to turn these machines into workspaces where assistants build and run workflows on their own.
- Some configurations are delayed: the 48GB P14 and 64GB P16 are due later this year, and pricing starts at C$3,499 for the P14 and C$4,999 for the P16 in Canada.
On October 7, 2026, ASUS opened pre-orders for a trio of machines, the ProArt P16 and P14 laptops and the GR1X mini PC, all built on NVIDIA's new RTX Spark platform, with a promise that serious AI work can now live on a desk or in a bag rather than in a distant data center. The pitch rests on memory, up to 128GB of it, and on the old human preference for owning one's tools instead of renting them by the use. The figures come from the makers themselves and await the verdict of independent testing, a reminder that every new machine arrives first as a promise. Shipping is expected from October 16, and the coming weeks will show how much of that promise survives contact with real work.
On October 7, 2026, ASUS opened global pre-orders for three computers built around one conviction: that serious AI work, including running autonomous software agents, can happen on a machine on your desk or in your bag. The lineup is the ProArt P16 and P14 laptops and the ProArt GR1X mini PC, all powered by NVIDIA's RTX Spark platform. The announcement came from Toronto and is a company press release, so every performance figure is a manufacturer's claim, not an independent result.
The centerpiece is memory. RTX Spark pairs a Blackwell RTX GPU with a Grace CPU and offers up to 1 petaflop of AI performance and 128GB of unified memory. ASUS argues that memory capacity is the chief limit of conventional PCs, and that this much lets larger language models, image and video pipelines, and complex 3D scenes run locally. It also concedes that what actually fits depends on model, quantization, context length and available memory.
The company's case is that owning this capacity replaces unpredictable pay-per-use cloud spending with fixed compute, keeps prompts and project files on the device, and lets users choose their own models. Cloud services remain valuable for frontier-scale work, ASUS says, and the machines support local, cloud or hybrid use.
Software completes the package. MuseTree combines FLUX and Wan models for token-free generation, StoryCube organizes media, and Adobe and Google bundles are offered. ASUS has partnered with Nous Research, whose open-source Hermes agent works across apps, terminal and local files with persistent memory, and ComfyUI, the visual workflow platform, is named as a foundation for agents to build on. Nous Research's Dillon Rolnick and Comfy's Yoland Yan both framed the pitch as intelligence people own rather than rent.
The hardware varies by form. The P16 is 12.9mm thick and 1.77kg with a 4K, 120Hz OLED panel; the P14 is the lightest ProArt yet at 1.48kg; the GR1X is a 150mm cube with dual fans, 10GbE and support for four 4K displays.
In Canada, the P14 starts at C$3,499 and the P16 at C$4,999. Some configurations, including the 48GB P14 and 64GB P16, arrive later this year. Shipping is expected from October 16, and the real test will be independent benchmarks and whether the promised configurations appear as scheduled.
On October 7, 2026, ASUS opened global pre-orders for three new computers built around a single idea: that serious AI work, including the running of autonomous software agents, can happen on a machine sitting on your desk or in your bag. The lineup is the ProArt P16 and ProArt P14 laptops and the ProArt GR1X mini PC, all powered by NVIDIA's new RTX Spark platform. The announcement went out from Toronto, and it is a company press release, so the performance figures below are ASUS and NVIDIA claims rather than independent test results.
RTX Spark pairs an NVIDIA Blackwell RTX GPU with an NVIDIA Grace CPU, linked by NVLink-C2C, and includes fifth-generation Tensor Cores with FP4 support. ASUS says the platform delivers up to 1 petaflop of AI performance and up to 128GB of unified memory. That memory ceiling is the center of the pitch: ASUS argues that memory capacity is one of the biggest limits of conventional PCs, and that this much of it lets larger language models, image and video generation pipelines, LoRA customization and complex 3D scenes run locally. It adds the caveat that what fits depends on model architecture, quantization, context length and available memory.
The company lists three advantages of owning the capacity. Local use replaces variable pay-per-use cloud spending with predictable compute, so people can regenerate and experiment without per-token or per-generation charges. Prompts, source files and project data stay on the device, and workflows can remain available offline. And users choose their own models and build repeatable workflows. ASUS also says cloud services stay valuable for frontier-scale models and burst capacity, and that the machines support local, cloud or hybrid execution.
Software is a large part of the package. ASUS preloads MuseTree, which combines the FLUX and Wan image and video models for token-free local generation, and StoryCube, which handles scene recognition and media organization. Adobe Creative Cloud bundles are offered, and eligible laptops get a Google AI bundle. On the agent side, ASUS has partnered with Nous Research, maker of the open-source Hermes agent, which works across desktop apps, the terminal and local files with persistent memory and isolated sandboxes. Through a dedicated Hermes Skill it connects to MuseTree. ASUS also names ComfyUI, the open-source visual workflow platform it says is used by millions, whose design lets assistants like Hermes build and run workflows, and its own assistant, Zenni Claw.
Nous Research CEO Dillon Rolnick said Hermes was built to provide intelligence people can own rather than rent, and that the machines have the memory to run powerful models locally and keep workflows on the user's own computer. Yoland Yan, CEO and co-founder of Comfy, said creators want to run the best models on their own machines and that ComfyUI makes workflows easy to share, reuse and hand to agents.
The hardware differs by form. The ProArt P16 is 12.9mm thick and 1.77kg, which ASUS calls its thinnest 16-inch ProArt, with a battery up to 99.9Wh. The P14 is 13.9mm and 1.48kg, the lightest ProArt laptop to date, with up to 90Wh. Both use CNC-machined bodies and Lumina Pro OLED displays; the P16 listing shows a 4K, 120Hz panel with up to 1600 nits HDR peak brightness. The GR1X measures 150 x 150 x 51mm and offers dual-fan cooling, dual M.2 SSD slots, 10GbE, WiFi 7, Bluetooth 5.4 and support for up to four 4K displays. It comes in silver and black.
Pricing was given for Canada. Pre-orders began October 7, with shipping estimated from October 16. The P14 starts at C$3,499 and the P16 at C$4,999. The P14 launches with 24GB or 128GB of LPDDR5X memory, the P16 with 32GB or 128GB; a 48GB P14 and 64GB P16 are due later this year. The ASUS Store carries the configurations, and Best Buy Canada will sell the 128GB P14 and 32GB P16. ASUS notes that configurations, pricing and availability may vary by region.
ASUS also placed the machines within a wider range that includes GeForce RTX 50 Series systems, the Ascent GX10 built on NVIDIA DGX Spark, and the ExpertCenter Pro ET900 built on DGX Station. What to watch is how the machines perform in independent testing after October 16, and whether the 48GB and 64GB models arrive as promised.
Citas Notables
We built Hermes to provide intelligence you can truly own, rather than simply rent.— Dillon Rolnick, CEO of Nous Research
Creators want to run the best AI models on their own machines, with full control.— Yoland Yan, CEO and co-founder of Comfy
Quieres la nota completa?
Lee el original en The Manila Times ↗
Technology