On an October day in 2026, ASUS put its ProArt P16 and P14 laptops on sale across Canada, machines built around NVIDIA's RTX Spark platform and designed to run artificial intelligence work on the device itself rather than in a distant data center. The move reflects a familiar pull in technology between renting capability from the cloud and owning it outright, a choice that trades variable monthly charges for fixed hardware. Yet the case for this shift arrives in the language of a vendor's own promotion, and its promised speed and savings have not yet met the test of independent measurement.
ASUS ProArt P16 and P14 Laptops with NVIDIA RTX Spark Begin Shipping in Canada
Intelligence you can own, rather than simply rent
Mark
So what actually changed on October 16?
Mimi
Two laptops, the ProArt P16 and P14, went on sale in Canada with RTX Spark inside, and select configurations started shipping that day.
Mark
Why does the memory figure get so much attention?
Mimi
Up to 128GB of unified memory is what ASUS says lets big models run locally instead of in the cloud.
Luke
Worth noting that's ASUS's framing. The 1 petaflop FP4 figure is a peak number, and the release itself says real results depend on model, quantization and context length.
Mark
And the cost argument?
Mimi
Local runs avoid per-token or per-generation fees, so heavy users could trade variable bills for a fixed purchase.
Luke
Fixed, but not small. C$3,499 and C$4,999 are starting prices, and the 128GB models cost more than that. No one has shown the break-even yet.
Mark
What about privacy?
Mimi
Data can stay on the device and work offline after setup, but only with fully local models and tools.
Luke
Right, and the release admits cloud-connected tools may send data off device. Nothing here is independently tested.
Mark
What should readers watch?
Mimi
Reviews, and whether the 48GB and 64GB models arrive later this year as promised.
Le Pouls
- ASUS has begun selling its ProArt P16 and P14 laptops across Canada, with some configurations shipping the same day they were announced.
- The pitch targets a costly habit: cloud AI services that bill for every token or generation, which ASUS wants buyers to replace with fixed local computing.
- Claims of up to 1 petaflop of AI performance and 128GB of unified memory remain unverified, and their real-world results depend on model size, quantization and memory limits.
- Partnerships with Nous Research, ComfyUI and others are knitting together local agents and creative tools, while cloud bundles stay available for the largest jobs.
- Independent reviews and actual street pricing will decide whether the local-first promise holds up once these machines reach working desks.
On an October day in 2026, ASUS put its ProArt P16 and P14 laptops on sale across Canada, machines built around NVIDIA's RTX Spark platform and designed to run artificial intelligence work on the device itself rather than in a distant data center. The move reflects a familiar pull in technology between renting capability from the cloud and owning it outright, a choice that trades variable monthly charges for fixed hardware. Yet the case for this shift arrives in the language of a vendor's own promotion, and its promised speed and savings have not yet met the test of independent measurement.
On October 16, 2026, ASUS announced from Toronto that its ProArt P16 and P14 laptops were on sale in Canada, with select configurations shipping the same day. Buyers can find them at the ASUS Store, Amazon Canada, Best Buy Canada, Memory Express and Newegg Canada. The announcement comes straight from the company and its partners, so the performance and savings claims have not been checked by anyone independent.
Both machines run on NVIDIA RTX Spark, which combines a Blackwell RTX GPU, a Grace CPU and fast interconnect technology. ASUS cites up to 1 petaflop of FP4 AI performance and up to 128GB of unified memory, enough, it says, to run large language and image models locally, though results will vary with model, memory and context length.
ASUS argues that local models avoid per-use cloud fees, keep prompts and project files on the device, and can work offline once set up. It also admits that cloud services still matter for the largest models and for sudden spikes in demand. The software leans on partners: MuseTree and StoryCube come preloaded, Hermes from Nous Research and ComfyUI workflows connect to the stack, and Adobe and Google bundles cover the cloud side.
The hardware is slim, with the P16 at 1.77kg and the P14 at 1.48kg, and both carry OLED screens. Prices start at C$3,499 for the P14 and C$4,999 for the P16, with larger memory options arriving later in 2026. ASUS also unveiled a small mini PC, the ProArt GR1X, for small teams. Whether the speed, savings and independence on offer hold up will become clear only when reviewers and buyers test them.
On October 16, 2026, ASUS announced from Toronto that its ProArt P16 and ProArt P14 laptops are on sale in Canada, with select configurations shipping the same day. Buyers can find them through the ASUS Store, Amazon Canada, Best Buy Canada, Memory Express and Newegg Canada. The announcement is a vendor press release, and the performance and savings claims in it come from ASUS and its partners rather than from independent testing.
Both machines run on NVIDIA RTX Spark, a platform that pairs an NVIDIA Blackwell RTX GPU with an NVIDIA Grace CPU, fifth-generation Tensor Cores with FP4 support, and NVLink-C2C interconnect technology. ASUS cites up to 1 petaflop of FP4 AI performance and up to 128GB of unified memory. The company says that memory capacity lets users run large language and vision-language models, image and video generation pipelines, LoRA customization and complex 3D scenes locally, though results depend on model architecture, quantization, context length and available memory.
ASUS frames the appeal in three points. Local models avoid per-token or per-generation cloud charges, turning variable spending into fixed computing capacity. Prompts, source files and project data can stay on the device, and fully local workflows can run offline after setup. And users pick their own models and build repeatable workflows. ASUS also concedes that cloud services still matter for frontier-scale models and burst capacity, and that cloud-connected tools may send data off the device and carry extra fees.
The software side leans on partnerships. ASUS preloads MuseTree, which integrates the FLUX and Wan models for local generation, and StoryCube, which handles scene recognition and media organization. Hermes, an open-source agent from Nous Research, can work across desktop apps, the terminal and local files, and connects to MuseTree through a dedicated skill. ASUS has also partnered with ComfyUI, the open-source visual workflow platform, whose workflows can be built and run by agents like Hermes. ASUS's own assistant, Zenni Claw, handles daily tasks and device control. Adobe Creative Cloud bundles and, on eligible laptops, a Google AI bundle cover the cloud side.
Dillon Rolnick, CEO of Nous Research, said Hermes is meant to offer intelligence users own rather than rent, with the memory to run powerful models locally. Yoland Yan, CEO and co-founder of Comfy, said creators want to run top models on their own machines with full control.
On hardware, the P16 is 12.9mm thick and weighs 1.77kg, which ASUS calls its thinnest 16-inch ProArt. The P14 is 13.9mm and 1.48kg. Batteries reach 99.9Wh in the P16 and 90Wh in the P14. Both use Lumina Pro OLED displays with Delta E under 1, up to 1,600 nits HDR peak brightness and 120Hz refresh. The P16 goes up to 4K with G-SYNC; the P14 up to 3K. Canada gets Nano Black P16 models and Nano Black or Neo White P14 models.
At launch the P14 comes with 24GB or 128GB of LPDDR5X memory, and the P16 with 32GB or 128GB. A 48GB P14 and a 64GB P16 are due later in 2026. The P14 starts at C$3,499 and the P16 at C$4,999. The listed P16 specifications show configurations with 18-core or 20-core Grace CPUs and 5,120 or 6,144 GPU cores.
ASUS also announced the ProArt GR1X, a 150 x 150 x 51mm mini PC with RTX Spark, 10GbE, WiFi 7 and support for four 4K displays, aimed at small teams. It sits in a wider lineup that includes GeForce RTX 50 Series systems, the Ascent GX10 built on NVIDIA DGX Spark, and the ExpertCenter Pro ET900N G3 built on DGX Station. Independent reviews and real-world pricing will show whether the claims hold up.
Citations marquantes
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
Envie de l'histoire complète ?
Lire l'original sur The North Bay Nugget ↗
Technology