$6,399 — that's how much GMKTec sells its EVO-X5 Pro mini PC with an AMD Max+ PRO 495 and 192GB RAM, and that's the cheapest fully loaded Gorgon Halo device

GMKtec EVO-X5 Pro
(Image credit: GMKtec)

  • GMKtec EVO-X5 Pro starts at $6,399, but other Gorgon Halo systems exceed $7,000
  • Runs 320B-parameter models offline under suitable conditions
  • Memory bandwidth reaches 273GB/s, and up to 160GB can serve graphics

GMKtec has launched the EVO-X5 Pro starting at $6,399, making it one of the earliest devices to use the Gorgon Halo chip.

The mini PC combines AMD’s Ryzen AI Max+ PRO 495 processor with 192GB of LPDDR5X-8533 unified memory for demanding local computing.

According to GMKtec, this device can run LLMs containing 320 billion parameters completely offline under suitable software and configuration conditions.

Latest Videos FromTechRadar

A $6,399 machine with substantial hardware

The company revealed that the starting price applies to the first 30 units of the 192GB memory and 2TB storage model.

After the first 30 units, subsequent early buyers will spend $200 more, purchasing this device at $6,599, while the regular price after the promotional period is $6,799.

A second model with 192GB of memory and 4TB of storage is available for $6,699 during the Super Early Bird promotion, $6,899 during the Early Bird period, and $7,099 at its regular price.

Other Gorgon Halo devices start well in excess of $7,000, making the EVO-X5 Pro one of the cheapest Gorgon Halo devices available.

However, the X5 Pro could easily be one of the most expensive mini PCs out there, as most of these devices sell for less than $4,000.

Its Ryzen AI Max+ PRO 495 processor provides 16 Zen 5 CPU cores, 32 threads, and boost speeds reaching 5.2GHz for demanding desktop workloads.

For graphics processing, this device uses the Radeon 8065S hardware with 40 compute units, RDNA 3.5 architecture, and hardware ray tracing.

Its XDNA 2 NPU can deliver up to 55 TOPS for supported artificial intelligence workloads, adding another processing resource to the system.

Memory bandwidth reaches 273GB/s, while the system can dynamically allocate as much as 160GB of its unified memory for graphics processing.

Its storage capacity can be expanded up to 24TB, giving the compact desktop considerable room for local datasets, applications and large artificial intelligence models.

To move that data in and out quickly, connectivity includes two USB4 V2 ports supporting up to 80Gbps, two USB4 ports reaching 40Gbps and two 10GbE LAN connections.

Offline AI capability is central to the EVO-X5 Pro

GMKtec originally developed the machine around support for 300B parameter models before later tuning increased its stated capability to 320B.

The company cautions that actual compatibility and performance depend on the selected model, software, quantization method, workload, and complete system configuration.

That capability is paired with features intended for private AI deployment, including local language models, coding assistants, research workloads and enterprise applications.

The EVO-X5 Pro supports eGPUs and multi-system clustering, allowing users to expand computing resources when individual workloads exceed its internal hardware.

For business environments, it includes AMD DASH, a dedicated dTPM 2.0 security chip and operating-system-independent out-of-band management.

GMKtec says the system has been validated for continuous 24/7 operation, although that claim does not establish performance levels across every workload.

To sustain that workload, cooling uses a vapor chamber with three fans inside an all-metal CNC-machined enclosure. The system also offers three performance modes for different operating requirements.

Beyond AI work, the device supports displays reaching 8K resolution, extending its intended use toward demanding professional desktop workloads.

We already have a unit of this mini PC on our desk; stay tuned for the detailed review coming up soon.


Google logo on a black background next to text reading 'Click to follow TechRadar'

Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds.

Efosa Udinmwen
Freelance Journalist

Efosa has been writing about technology for over 7 years, initially driven by curiosity but now fueled by a strong passion for the field. He holds both a Master's and a PhD in sciences, which provided him with a solid foundation in analytical thinking.

You must confirm your public display name before commenting

Please logout and then login again, you will then be prompted to enter your display name.