The GMKtec EVO-X5 Pro has emerged as a formidable mini PC designed for high-end professional workloads and local artificial intelligence applications. Powered by the AMD Ryzen AI Max+ PRO 495 processor and equipped with Radeon 8065S graphics, the device stands out primarily due to its massive 192GB of unified LPDDR5X memory. This configuration allows the machine to run large language models (LLMs) locally, including those with up to 300 billion parameters, without stalling, a capability that distinguishes it from smaller competitors that often hit memory limits quickly.

In terms of design, the EVO-X5 Pro occupies a middle ground between traditional mini PCs and small tower desktops. It features an understated all-metal black finish with dimensions of 221 x 80 x 195mm and weighs 2.5kg. The chassis includes a fingerprint-recognizing green power button, a P-Mode switch for adjusting power consumption between 54W and 120W, and a comprehensive array of ports. Connectivity options include two USB4 v2 ports capable of 80Gbps speeds, dual 10GbE RJ-45 DASH network ports, HDMI 2.1, and Wi-Fi 7. The internal layout supports up to three M.2 SSD slots, allowing for a total storage capacity of up to 24TB, though the RAM is soldered and non-upgradeable.

Performance testing revealed that the EVO-X5 Pro handles standard Office applications and creative suites like Adobe Photoshop and Lightroom with exceptional speed. In real-world scenarios involving 4K video editing from a Canon EOS R5C, the machine performed admirably even without discrete graphics, leveraging its unified memory architecture. While scrolling through complex timelines in DaVinci Resolve showed some slowdowns compared to systems with dedicated GPUs, the performance remained comparable to traditional workstations. The inclusion of dual 10GbE ports enabled efficient network video editing by accessing files directly from a NAS, eliminating the need for local desktop storage during the workflow.

The device's primary strength lies in its AI capabilities. Pre-installed software like GMKtec Claw and Herdsman facilitates local model management and inference. During testing, the system successfully ran the DeepSeek-V4-Flash-0731 model (155.7GB), though generation speeds were slow due to limited headroom. Switching to a smaller Qwen3 30B-A3B model resulted in significantly faster and more practical response times. The manufacturer claims the 55 TOPS NPU, combined with the memory boost, supports running LLMs up to 320 billion parameters locally.

Despite its impressive specifications and performance, the EVO-X5 Pro comes at a premium cost. It is available now for $6,799 in the US or £5,299 in the UK, with an additional $400/£300 required for the 4TB storage version over the base 2TB model. While the machine offers a stylish design and powerful creative workstation capabilities, its high price point may be prohibitive for users who do not require its specific AI processing power or massive memory capacity. For those building personal data centers or requiring offline AI workflows, however, it represents a significant leap in mini PC performance.

Sources: TechRadar.