GMKtec has officially launched its EVO X5 Pro AI workstation, pairing AMD's Ryzen AI Max Plus PRO 495 processor with 192GB of unified memory and enough local memory capacity to run large AI models with up to 320 billion parameters.
The flagship configuration starts at $6,599 with 192GB of LPDDR5X memory and 2TB of storage. A second model with the same memory capacity and 4TB of storage is listed at $6,899.
GMKtec originally showed the system at IFA 2026 in Berlin, where it demonstrated the workstation's local AI capabilities. The retail launch now brings the machine into direct competition with other compact systems based on AMD's latest high end Ryzen AI platform.
Ryzen AI Max Plus PRO 495 handles CPU, GPU and AI workloads
The EVO X5 Pro uses the Ryzen AI Max Plus PRO 495, a Zen 5 based processor with 16 CPU cores and 32 threads.
The chip can boost up to 5.2GHz and includes Radeon 8065S integrated graphics based on AMD's RDNA 3.5 architecture. The GPU contains 40 Compute Units.
GMKtec also highlights the processor's AI capabilities, with the platform rated at up to 131 AI TOPS.
| Specification | GMKtec EVO X5 Pro |
|---|---|
| Processor | AMD Ryzen AI Max Plus PRO 495 |
| CPU | 16 cores, 32 threads |
| Maximum boost clock | Up to 5.2GHz |
| Integrated GPU | Radeon 8065S |
| GPU architecture | RDNA 3.5 |
| GPU Compute Units | 40 |
| AI performance | Up to 131 TOPS |
| Memory | 192GB LPDDR5X unified memory |
| Maximum VRAM allocation | Up to 160GB |
| Included storage | 2TB or 4TB |
| Maximum storage | Up to 24TB |
| Ethernet | Dual 10GbE |
| Starting price | $6,599 |
The large unified memory pool is central to the workstation's AI focus.
Out of the 192GB available, as much as 160GB can reportedly be assigned as graphics memory. That allows the integrated GPU to work with AI models that would normally require large amounts of dedicated VRAM.
GMKtec says the system can now run 320B parameter models
The EVO X5 Pro was initially designed around local models with roughly 300 billion parameters.
GMKtec now says software and hardware optimizations have pushed that capability further, allowing the workstation to handle models with as many as 320 billion parameters locally.
That figure represents the company's stated capability and will depend on factors including model format, quantization, memory requirements and workload configuration.
Still, the large unified memory architecture is what makes workloads of this scale possible on a single compact system without relying on several discrete GPUs.
The design follows a broader trend among Ryzen AI Max Plus systems, where CPU, GPU and NPU resources share a large pool of fast memory instead of separating system RAM from dedicated graphics memory.
Three M.2 slots allow up to 24TB of storage
Storage expandability is another major part of the EVO X5 Pro.
GMKtec includes three M.2 2280 SSD slots. The company says the system can support up to 24TB in total by installing three 8TB drives.
The entry configuration includes 2TB from the factory, while the more expensive version increases that to 4TB.
Buyers who plan to install their own SSDs can therefore choose the 2TB version and use the remaining expansion capacity later.
This level of storage can be particularly useful for local AI workloads because large models, datasets and multiple quantized variants can consume hundreds of gigabytes.
USB4 and dual 10GbE provide high speed connectivity
The EVO X5 Pro includes a wide selection of high speed connections.
It provides two USB4 ports rated at 40Gbps and another two USB4 v2 connections capable of up to 80Gbps with DisplayPort Alt Mode support.
Four USB Type A ports are also included.
For networking, GMKtec has fitted dual 10GbE LAN connections, giving the workstation enough bandwidth for fast local networking, shared storage or clustered configurations.
The system also supports AMD DASH remote management, which is aimed more toward professional and enterprise environments than conventional consumer mini PCs.
Pricing starts below $6,600
The 192GB and 2TB model is currently priced at $6,599, compared with a listed regular price of $6,799.
The 192GB and 4TB configuration costs $6,899.
That still places the EVO X5 Pro firmly in professional workstation territory, but its pricing is positioned below some competing AI systems offering similar memory capacity and Ryzen AI Max Plus hardware.
The main appeal is the combination of 192GB unified memory, Radeon 8065S graphics, a 16 core Zen 5 CPU and the ability to allocate up to 160GB of memory to GPU workloads.
For developers and researchers interested in running very large AI models locally, the EVO X5 Pro offers a compact alternative to traditional multi GPU workstation setups, while also providing substantial storage expansion and high speed networking.
Discussion (0)
Be the first to comment.