A Mini PC 128GB RAM Platform That Behaves Like a Workstation
The GMKtec EVO-X2 is an AMD Ryzen AI Max mini PC that combines 128GB of system RAM, up to 96GB of shared VRAM, and a compact chassis to deliver workstation-class performance for AI, graphics, and 4K media workloads while staying on the desk instead of under it.
GMKtec has added a new EVO-X2 variant that keeps the headline feature—128GB RAM—but trims default storage to make the configuration more affordable. Rather than chasing raw capacity at any cost, this release focuses on cost-optimising a machine that was already aimed at ultra-high-end users. The system is still built around AMD’s Strix Halo platform and the Ryzen AI Max+ 395 processor, so buyers do not sacrifice CPU or GPU capability to reach the lower price band. This is the key point: the new EVO-X2 is not a cut-down toy; it is the same compact workstation graphics platform with smarter compromises where they hurt least.
Why 128GB RAM and 96GB VRAM Matter for Compact Workstation Graphics
The defining move here is memory, not storage. With 128GB of system RAM, the EVO-X2 can allocate up to 96GB as VRAM to its integrated Radeon 8060S graphics engine. That number is not a marketing flourish; it is on par with, or beyond, what many large discrete GPUs offer, but delivered in a mini PC 128GB RAM platform that fits on a bookshelf. "With 128GB RAM and up to 96GB VRAM allocation, the EVO-X2 turns an integrated GPU system into a practical compact workstation graphics solution."
This shared memory design changes what an iGPU machine can do. Tasks like training and running large language models locally, displaying high-resolution video, creating complex 3D models, and processing large scientific datasets move from theoretical to feasible workloads on a desk-friendly box. Instead of treating mini PCs as light-duty companions, GMKtec is arguing that they can replace mid-range towers for serious creators and AI developers—provided you feed them enough RAM and a capable graphics core.
Ryzen AI Max+ 395: Local AI Computing Hardware Without the Cloud
At the heart of this AMD Ryzen AI Max mini PC is the Ryzen AI Max+ 395, built on AMD’s latest Zen 5 architecture and paired with the Radeon 8060S iGPU. This is not a generic laptop chip dropped into a small case; it is part of the Strix Halo platform, designed for high multi-core CPU throughput and strong integrated graphics inside compact systems. Crucially, the chip’s AI compute capabilities are meant for local AI computing hardware scenarios, not as a side feature.
Real-world tests from earlier EVO-X2 releases showed the 395 can hold peak performance for long periods without thermal throttling. That stability matters more than flashy benchmark spikes when you are running multi-hour 3D renders, fine-tuning language models, or crunching through large data pipelines. Combined with up to 96GB VRAM allocation, the system enables AI model inference and even training runs to stay entirely offline, avoiding cloud latency, recurring subscription costs, and data privacy concerns.
The New Budget-Friendly Variant: Smarter Cuts, Not Crippled Hardware
GMKtec’s most important decision is where to save money: storage, not memory or compute. The new EVO-X2 variant pairs 128GB RAM with a 1TB SSD, instead of the previous 2TB default. This single change reduces the list price by USD 150 (approx. RM700) compared with the earlier 128GB+2TB configuration, which sits at around USD 3,649 (approx. RM17,000). In other words, the budget-friendly variant expands access to the high-performance mini PC segment by cutting the easiest component to upgrade later.
There is still a steep premium for memory: moving from 64GB to 128GB RAM adds about USD 1,500 (approx. RM7,000) to the EVO-X2’s price. That is painful if you only browse and stream, but it makes sense for professionals who need very high VRAM for deep learning algorithms or massive datasets. For that audience, disk capacity is secondary. They care that the platform can assign 96GB VRAM to the iGPU, run for hours without throttling, and sit unobtrusively on a desk. GMKtec’s configuration strategy reflects that reality.
Who Should Care: From 4K Editors to AI Tinkerers
This EVO-X2 update is unnecessary overkill for everyday users—and that is the point. It targets people who treat local AI computing hardware and GPU memory as their main bottlenecks, not their electricity bill. With VRAM on par with luxury discrete cards, the system can run local LLMs, handle 4K video editing, and deliver complex 3D rendering in a single compact box.
The practical impact is clear: creators and researchers can swap noisy towers and multi-GPU rigs for a desk-friendly mini PC that behaves like a small workstation. Instead of renting cloud GPUs, they can keep sensitive footage and datasets on site, control their environment, and still hit professional deadlines. GMKtec’s move does not democratise high-end compute in the sense of making it cheap, but it does make ultra-high-memory mini PCs more reachable and more sensible, especially for professionals who value space, silence, and ownership of their tools.
