A new mini PC can now run some of the largest artificial intelligence models entirely on your desk, without sending data to the cloud. GMKtec has unveiled the EVO-X5 PRO, a compact workstation powered by AMD's latest Ryzen AI MAX+ PRO 495 processor, which combines a traditional CPU, graphics processor, and specialized neural processing unit (NPU) on a single chip. The system can handle AI models with 300 billion parameters or more, a scale previously reserved for data centers and cloud services.

The shift toward running massive AI models locally represents a significant change in how professionals and researchers approach artificial intelligence work. Instead of uploading data to remote servers, users can now process complex AI tasks on hardware that sits under their desk. This approach offers privacy benefits, eliminates cloud computing costs, and removes latency delays that come from sending information across the internet.

What Makes This Mini PC Different From Cloud-Based AI?

The EVO-X5 PRO achieves this capability through a combination of hardware choices. The Ryzen AI MAX+ PRO 495 processor includes 16 CPU cores running at speeds up to 5.2 gigahertz, a graphics processor with 40 compute units, and an XDNA 2 neural processing unit that delivers up to 55 trillion operations per second (TOPS) of AI-specific computing power. Together, these components provide 131 TOPS of total performance.

The real enabler, however, is memory. The EVO-X5 PRO comes equipped with 192 gigabytes of unified memory, a type of memory architecture that allows the CPU, GPU, and NPU to access the same data pool without copying information between different memory systems. This eliminates a major bottleneck in AI processing. The system can dedicate up to 160 gigabytes of this memory to running large language models, making it possible to load and run models with 300 billion or more parameters locally.

GMKtec claims the EVO-X5 PRO delivers 66 percent faster CPU orchestration and 40 percent lower cost per completed AI workflow compared to NVIDIA's DGX Spark system, a competing solution designed for similar workloads. These performance and cost advantages could make local AI processing more attractive to professionals who currently rely on cloud services or expensive data center hardware.

How to Set Up Local AI Processing on the EVO-X5 PRO?

• Memory Configuration: The system supports up to 192 gigabytes of unified memory operating at 8533 megatransfers per second, allowing you to allocate up to 160 gigabytes specifically for GPU-based AI model execution without performance degradation.
• Storage Expansion: The EVO-X5 PRO includes three M.2 2280 PCIe Gen4 slots, each supporting up to 8 terabytes, for a total of 24 terabytes of storage capacity to house multiple large AI models and datasets locally.
• Power Management: Three selectable power modes allow users to switch between performance profiles depending on their AI task requirements, with the chip operating between 45 and 120 watts depending on the mode selected.
• Connectivity Options: Dual USB4 V2 ports on the rear support external graphics processors, while dual 10-gigabit Ethernet ports enable clustering multiple EVO-X5 PRO units together for distributed AI processing workloads.

The mini PC also includes software support through GMKtec's Claw AI agent solution, which helps manage and optimize AI workloads on the hardware. The system supports AMD Dash Remote Management technology, allowing administrators to monitor and control the device remotely through a dedicated network adapter. A dedicated TPM 2.0 security chip provides hardware-level encryption and security features for protecting sensitive AI models and data.

When Will This Hardware Be Available?

GMKtec has not yet announced pricing for the EVO-X5 PRO, but the company plans to launch the system in October 2026. The timing positions the device to capture growing interest in local AI processing, particularly among professionals who work with large language models, researchers developing AI applications, and enterprises seeking to reduce cloud computing expenses.

The underlying AMD processor, the Ryzen AI MAX+ PRO 495, represents the company's latest effort to integrate AI acceleration directly into consumer and professional-grade processors. By combining CPU, GPU, and NPU capabilities on a single chip with substantial unified memory support, AMD is enabling a new category of compact AI workstations that can handle tasks previously requiring specialized data center infrastructure.

This development reflects a broader industry trend toward edge AI processing, where computation happens locally rather than in distant cloud data centers. As AI models grow larger and more capable, the ability to run them without cloud dependency becomes increasingly valuable for privacy-conscious organizations, cost-sensitive operations, and users working with sensitive or proprietary data that cannot leave their premises.