01
What ASUS has announced
ASUS first unveiled RTX Spark-powered ProArt P16 and P14 laptops and a ProArt Mini PC at Computex in June 2026. On September 2, 2026, ASUS showcased the lineup at IFA under the P16, P14 and GR1X names with additional workflow and design detail.
That distinction matters for release timing: September 2 is the current IFA showcase and GR1X naming milestone, not the first public reveal of the RTX Spark ProArt concept.
02
What NVIDIA RTX Spark brings
ASUS describes RTX Spark as combining a Blackwell RTX GPU with 6,144 CUDA cores, a Grace CPU with up to 20 cores, fifth-generation Tensor Cores with FP4 support and NVLink-C2C between CPU and GPU.
ASUS rates the platform at up to 1 petaflop of AI performance and up to 128GB of high-bandwidth unified memory. It also says the platform can handle LLMs up to 120B parameters, 90GB-plus 3D scenes and 4K AI video generation. These are manufacturer/platform claims, not independent benchmarks.
03
ProArt P16 vs P14 vs GR1X
The ProArt P16 is the larger mobile creator system. ASUS lists it at 12.9mm and 1.77kg with a 99Wh battery, plus display options up to a 4K 120Hz Lumina Pro OLED panel.
The ProArt P14 is the more portable laptop at 13.9mm and 1.48kg with a 90Wh battery and display resolution up to 3K. Both use the RTX Spark platform.
The ProArt GR1X is a 150 × 150 × 51mm mini PC aimed at sustained and always-on workloads. ASUS lists 10GbE, Wi-Fi 7, Bluetooth 5.4, dual-fan cooling and support for up to four 4K displays.
04
Why unified memory matters for local AI
A large unified-memory pool can make model sizes and creative graphs practical that are difficult on conventional mobile GPUs with smaller dedicated VRAM pools. It also lets compatible workflows keep prompts and assets on-device instead of relying exclusively on cloud inference.
A model fitting in memory does not guarantee a particular speed, context length or quality. Precision, quantization, KV cache, runtime overhead and software optimization all affect real-world performance.
05
Local generative AI and creator software
ASUS is pairing the hardware with its creator apps and the CUDA ecosystem. MuseTree integrates local FLUX and WAN workflows, while StoryCube focuses on AI-assisted media organization and scene recognition.
ASUS is also demonstrating ComfyUI workflows and agent-assisted interactions between its Zenni Claw system and MuseTree. These demonstrations establish intended workflows, but exact performance depends on models, nodes and runtime support.
06
Price and availability
ASUS's earlier Computex material said availability would begin in fall 2026 in select regions, with more configuration and regional details to follow. The September 2 IFA release does not publish a global price list, preorder schedule or complete country-by-country retail matrix.
Regional rollout can differ. For example, ASUS previously said the P16 and P14 would arrive in Malaysia in Q3 2026 while mini-PC availability would be announced later. Buyers should verify current regional product pages rather than infer worldwide availability from a showcase announcement.
07
Who these systems are for
The strongest fit is creators and developers who repeatedly run local image generation, ComfyUI graphs, local LLM inference, coding agents, 3D work or private media and document workflows where memory capacity and local processing matter.
For occasional generation or frontier-model use, cloud services can remain simpler. A hybrid approach is more realistic: keep compatible recurring workloads local and use hosted APIs when a task needs models or scale the machine cannot provide.
08
Important launch limitations
ASUS has not yet provided a complete global price list, final regional SKU matrix or independent model-level inference benchmarks in the primary September 2 release. Treat unofficial pricing cautiously until first-party regional listings appear.
The 1-petaflop, 120B-model and 4K AI-video figures are ASUS/NVIDIA capability claims. They should not be read as guaranteed performance for every model, precision, context length or workflow.
Sources
Primary and supporting sources
Facts were rechecked against the linked sources immediately before publication. Pricing, product availability and rollout status can change.