Acer Veriton RA110: A Six-Inch AI Mini Workstation With Ryzen AI Max+ 395

Acer Veriton RA110: A Tiny AI Mini Workstation With Big-Iron Ambitions

Every so often a spec sheet lands on my desk that makes me read it twice, then pull out a tape measure. The Acer Veriton RA110 is that machine. Unveiled on May 29, 2026 as part of a broader Veriton workstation and desktop lineup, this AI mini workstation crams an AMD Ryzen AI Max+ 395 into a chassis that measures roughly six inches square and sits under two inches tall. The primary keyword here is unavoidable: this is a genuine AI mini workstation, and it is one of the most exciting small-form-factor reveals I have studied in a while.

Let me walk you through the numbers, because the numbers are the whole story.

The Ryzen AI Max+ 395 At The Heart Of It

The headline silicon is AMD's Ryzen AI Max+ 395, a 16-core, 32-thread part that pairs serious CPU muscle with a 40-core Radeon 8060S integrated GPU. That GPU core count is what made me sit up. We are not talking about a token iGPU bolted on for desktop duty; we are talking about a wide graphics array sharing the same package as the CPU, which is exactly the architecture you want when local AI inference is the goal.

Acer rates the platform at up to 126 TOPS of AI performance. For context, that figure comfortably clears the bar for a Windows 11 Copilot+ PC, and it is the kind of throughput that turns on-device acceleration from a marketing checkbox into a practical daily tool.

Running Local LLMs Up To 200 Billion Parameters

Here is the spec that genuinely impressed me. Acer says the Veriton RA110 can run local large language models up to 200 billion parameters. Read that again. A box you could hide behind a coffee mug is positioned to host models that, not long ago, lived exclusively in racked servers.

The enabler is memory. The RA110 offers up to 128GB of quad-channel LPDDR5X-8533 unified memory. Unified is the operative word: CPU and GPU draw from the same fast pool, so large model weights do not have to ping-pong across a narrow bus. Quad-channel LPDDR5X at 8533 MT/s is a lot of bandwidth to feed those 40 Radeon cores, and it is precisely why a 200-billion-parameter local LLM is on the table at all. Pair that with up to 2TB of M.2 2280 SSD storage and you have room for a respectable library of model checkpoints.

Six Inches Square: The Form Factor

I love a good measurement, so let me indulge. The chassis comes in at roughly 160 x 160 x 47mm. That is six inches square and under two inches tall, which puts the RA110 firmly in mini-PC territory while delivering workstation-class AI compute. The density-to-performance ratio here is the part that earns my hardware-reviewer enthusiasm. Fitting a 16-core processor, a 40-core GPU, and 128GB of unified memory into that envelope is a thermal and engineering achievement worth applauding.

Connectivity And The Acer Sense Pro Dashboard

Acer did not skimp on the ports, which I always appreciate in a desktop-class machine. The RA110 includes WiFi 7, Bluetooth 5.4, two USB4 Type-C ports, 2.5GbE wired networking, and HDMI 2.1. That is a forward-looking I/O stack: USB4 for high-bandwidth peripherals and external displays, 2.5-gigabit Ethernet for moving model files quickly, and WiFi 7 for when you cannot run a cable.

The software touch I find most thoughtful is the Acer Sense Pro dashboard. Rather than leaving you to guess at inference performance, it surfaces the two metrics that actually matter when you run local models: Tokens-per-Second and Time-to-First-Token. As someone who lives for benchmarks, having those numbers exposed natively is a small but meaningful quality-of-life win. It treats the user like an enthusiast who wants real data, not a spinner.

Why This Mini Workstation Matters

Now for the analysis, clearly labeled as my opinion. The Veriton RA110 reads like a deliberate answer to a question more people keep asking: can I run capable AI locally, privately, on my own desk, without a server rack? Based on these confirmed specs, the answer is shaping up to be a confident yes. The combination of 126 TOPS, 128GB of unified memory, and that 40-core Radeon 8060S is purpose-built for exactly that workload.

I want to stress that I have not run my own tests, so real Tokens-per-Second figures will have to wait. But the architecture is sound and the ambition is refreshing.

Availability

Acer plans to offer the Veriton RA110 in select markets, including North America, in the second half of 2026. I will be first in line to put a stopwatch on that Time-to-First-Token metric. Until then, this six-inch AI mini workstation has earned a permanent spot on my watchlist.

Sources: Liliputing — Acer Veriton RA110 is a mini workstation with Ryzen AI Max+ 395, May 29, 2026; PR Newswire (Acer) — Acer Unveils Comprehensive Lineup of Veriton Workstations and Desktops, May 29, 2026; Profesional Review — Acer Veriton RA110 AI mini workstation, May 2026