The ONEXStation Mini AI PC Hits Pre-Order — Ryzen AI Max+ 395, 128GB Unified Memory, $2,999

A Strix Halo Desktop Mini PC Built for Local Large-Model AI

ONEXPLAYER opened pre-orders for the ONEXStation Mini AI PC on May 18, 2026 — a compact aluminum-chassis desktop mini PC built around AMD's Ryzen AI Max+ 395 "Strix Halo" APU, paired with 128GB of LPDDR5x-8000 quad-channel unified memory and positioned squarely at the local large-model AI workload audience. The headline configuration with Ryzen AI Max+ 395, 128GB memory, and 1TB NVMe storage is listed at $2,999 (down from a stated $3,599 reference price), with a keyboard and mouse bundle. For the mini PC and self-hosted LLM community, this is one of the most aggressively-specced local AI desktops to land in 2026 and the cleanest demonstration yet of what the Strix Halo platform can do when it has full desktop thermal headroom to work with.

For anyone tracking how the Ryzen AI Max+ 395 single-board mini PC platform is maturing into a credible local AI workstation, the ONEXStation is the launch to watch. The Strix Halo APU has been the most-discussed AMD mobile silicon of the year in the mini PC community thanks to its 16 Zen 5 cores, integrated Radeon 8060S graphics with 40 RDNA 3.5 compute units, and the 256GB/s memory bandwidth that the LPDDR5x-8000 quad-channel configuration unlocks. Putting that silicon in a desktop chassis with room for sustained cooling is the structural step that lets the platform run local inference workloads up to 100 billion parameters — the kind of capability that previously demanded a much larger and louder workstation.

What the ONEXStation Brings to the Mini AI PC Category

The structural pitch is local-large-model AI in a desktop mini PC chassis. The ONEXStation packs the Ryzen AI Max+ 395 — 16 Zen 5 CPU cores, 32 threads, boost clocks up to 5.1 GHz, and the Radeon 8060S integrated GPU — into a 186 x 193 x 62mm aluminum chassis (roughly 7.3" x 7.6" x 2.4"). The 128GB of LPDDR5x-8000 unified memory is the headline specification, because up to 96GB of that can be allocated as VRAM, giving local AI workloads the kind of memory headroom that GPU-based workstations have historically required dedicated enterprise hardware to match.

Why the Unified Memory Architecture Matters for Local AI

The single most important technical detail in the ONEXStation specification is the unified memory model. On a traditional GPU-based AI workstation, the model weights have to fit in the GPU's dedicated VRAM — and that constraint has historically meant either expensive multi-GPU configurations or running smaller quantized models. The Strix Halo unified memory architecture means the same memory pool serves the CPU and the integrated GPU. Local inference workloads up to 100B parameters fit in that 128GB pool, which is the structural difference between a desktop AI workstation that can host modern frontier-scale local models and one that cannot.

The Storage, Connectivity, and Cooling Story

Two M.2 2280 slots support PCIe 4.0 x4 NVMe storage, giving builders the room to mix a fast OS drive with a large model-cache drive. WiFi 7 with Bluetooth 5.4 handles wireless connectivity, and the chassis includes the standard mini PC port complement for displays, USB peripherals, and Ethernet. The aluminum chassis design pairs the thermal headroom required by Strix Halo's 120W power envelope with the kind of build quality that the mini PC community expects from the higher end of the category.

What the $2,999 Price Tier Buys You

At $2,999 for the headline configuration, the ONEXStation is positioned at the premium end of the mini PC market but well below the cost of a comparable GPU-based AI workstation that could run 100B-parameter models locally. For developers building local AI applications, researchers running inference experiments outside the major cloud APIs, and self-hosted LLM enthusiasts who want frontier-scale local models without renting a cloud GPU, the pricing math is one of the cleanest local-AI value propositions on the market.

How the ONEXStation Lands Against the Strix Halo Mini PC Category

The Ryzen AI Max+ 395 platform has been arriving across multiple form factors throughout 2026 — handhelds like the ONEXPLAYER X2 Mini, larger desktop mini PCs from competing brands, and the variety of all-in-one and AIO-PC variants that have shown up at CES, Computex, and other industry events. The ONEXStation sits in the dedicated desktop mini PC tier, which is the right form factor for the local AI workload audience because it removes the thermal compromises that handheld and ultraportable form factors impose on sustained inference performance.

Why Dedicated Desktop Cooling Unlocks the Platform

The structural reason a dedicated desktop chassis matters for Strix Halo is that the platform's peak performance is gated by sustained thermals. Handheld and ultraportable form factors throttle the APU once thermal headroom runs out, which is acceptable for gaming workloads but problematic for long-running local AI inference jobs. The ONEXStation desktop chassis lets the platform run at its full performance envelope for hours at a time, which is the structural difference that makes the unit a credible local AI workstation rather than just a portable inference experiment.

The Setup Going Forward

For mini PC enthusiasts, self-hosted LLM builders, and the broader local AI community, the May 18 ONEXStation pre-order opening is one of the cleanest entries into the Ryzen AI Max+ 395 desktop platform to date. The 128GB unified memory specification gives local AI workloads the headroom they need. The aluminum desktop chassis gives Strix Halo the thermal envelope it needs. The $2,999 pricing tier brings frontier-scale local inference within reach of serious developers and enthusiasts. The next watch items are the global shipping cadence, independent benchmark coverage of the Strix Halo platform at desktop thermal envelopes, and how competing mini PC brands respond with their own Ryzen AI Max+ 395 desktop configurations. For anyone tracking the local AI workstation category, the ONEXStation is the configuration to evaluate.

Sources: VideoCardz, "ONEXStation announced, Mini AI PC with Ryzen AI Max+ 395 for $2,999," May 2026; Liliputing ONEXStation coverage, May 2026; WCCFTech ONEXStation Mini AI PC launch, May 2026; MiniXPC ONEXStation product listing, May 18, 2026; Guru3D ONEXStation chassis coverage, May 2026; XiaomiToday ONEXStation specs, May 2026.