Tom's Hardware and WCCFTech report that GMKtec has unveiled the dramatically redesigned EVO-X3, a flagship AI mini PC workstation built around AMD's Ryzen AI Max+ 395 'Strix Halo' processor (16-core Zen 5). WCCFTech says the unit packs 128 GB of LPDDR5X unified memory, an OCuLink (PCIe Gen4 x4) port for eGPU expansion, triple-fan cooling in a PS4-sized chassis, and GMKtec's Claw+Wrangler local-inference AI suite. Tom's Hardware notes the chassis bears a visible signature from AMD CEO Lisa Su. Early access opens June 22 and global launch is June 29, priced at $3,600 (128 GB / 2 TB) and $3,849 (128 GB / 4 TB). A Ryzen AI MAX+ 495 variant with 192 GB memory is planned for later in 2026.
What happened
Tom's Hardware and WCCFTech report that GMKtec has unveiled the EVO-X3, a dramatically redesigned flagship AI mini PC workstation built around AMD's Ryzen AI Max+ 395 'Strix Halo' processor (16-core Zen 5, TSMC 4nm). The unit ships with 128 GB of LPDDR5X unified memory at 8000 MT/s and dual PCIe 4.0 storage bays supporting up to 4 TB. Tom's Hardware notes the chassis bears a visible signature from AMD CEO Lisa Su. WCCFTech reports early access opens June 22 and the global launch is June 29, priced at $3,600 (128 GB / 2 TB) and $3,849 (128 GB / 4 TB). A Ryzen AI MAX+ 495 variant with 192 GB memory is planned for later in 2026.
Key hardware upgrades over EVO-X2
WCCFTech identifies the most important addition as an OCuLink (PCIe Gen4 x4) port, enabling much faster eGPU support than the previous USB4 interface. The new chassis is taller to accommodate triple-fan cooling with 41 mm metal-bladed fans and a sustained 140 W thermal envelope (54 W Balanced / 80 W Performance modes). The platform also gains Wi-Fi 7 and retains USB4 ports. Despite being taller, the overall footprint remains PS4-sized.
AI software
GMKtec ships the EVO-X3 with its Claw+Wrangler AI suite for 24/7 local agent and inference capabilities, requiring no token costs or cloud dependency, per WCCFTech.
Competitive context
WCCFTech frames the EVO-X3 as a direct rival to NVIDIA's DGX Spark at a lower price point. AMD's unified memory architecture lets the Radeon 8040S GPU access the full 128 GB pool, enabling larger local models than discrete GPU setups at the same memory capacity. Independent inference throughput benchmarks against NVIDIA DGX Spark and AMD's Ryzen AI Halo reference design have not yet been published.
What to watch
- •Third-party thermal, noise, and inference benchmarks under sustained AI workloads.
- •Performance comparisons against NVIDIA DGX Spark and AMD Ryzen AI Halo.
- •Availability and pricing of the 192 GB Ryzen AI MAX+ 495 variant expected later in 2026.
Scoring Rationale #
Competitively priced pre-assembled mini workstation targeting local AI inference practitioners, notable for OCuLink eGPU support and 128 GB unified memory at $3,600. Relevant but narrow hardware product release from a smaller-brand vendor; no independent benchmarks yet published.
Practice interview problems based on real data
1,500+ SQL & Python problems across 15 industry datasets — the exact type of data you work with.