At IFA 2026 in Berlin, ACEMAGIC has placed before the world a small but consequential object: a 2-liter workstation housing AMD's new Ryzen AI Max+ PRO 495 processor, capable of running vast artificial intelligence models entirely on-device. It is a quiet challenge to the assumption that serious computation must live in the cloud or in a tower, and a signal that the boundary between portable and powerful is being redrawn. For developers, researchers, and creators who handle sensitive data or simply wish to work untethered, this machine arrives as both a practical tool and a philosophical propo
ACEMAGIC Debuts Ryzen AI Max+ PRO 495 Mini Workstation at IFA 2026
A machine that fits in your hand but thinks like a workstation
Mark
So this is a tiny computer that can run a 284-billion-parameter AI model. How is that possible?
Mimi
The unified memory is the key. Normally, a CPU, GPU, and AI accelerator each have their own memory pools, and moving data between them is slow. Here, they all share the same 192 gigabytes. The data doesn't have to move around.
Luke
But we should be careful about what "running" means. ACEMAGIC says it's demonstrated the model running on the device. That's not the same as saying it runs at production speed, or that every use case works smoothly.
Mimi
Fair point. The source says actual performance varies depending on the model, quantization method, software, and system configuration. It's not a blanket claim.
Mark
Why would someone want to run AI locally instead of using the cloud?
Mimi
Privacy, for one. If you're processing sensitive data—medical records, proprietary designs, classified information—you don't want it leaving your machine. Speed is another. No network latency. And cost, if you're doing this work constantly.
Luke
Those are all real reasons. But the source doesn't give us pricing, power consumption, or thermal performance. We don't know if this thing runs hot, or what it costs compared to a cloud subscription.
Mark
What's the actual market here?
Mimi
Developers testing AI models, content creators using AI tools for video or image work, researchers who can't send data to the cloud. Anyone who needs portable computing power without cloud dependency.
Luke
The source says it's intended for those groups, but we don't have evidence yet that they're actually buying it. This is a product announcement at a trade show.
Mark
So we're watching the beginning of something, not the end.
Mimi
Exactly. ACEMAGIC is among the first to ship this processor in a consumer-facing device. That's the news—the timing, the form factor, the capability. What happens next is still open.
Le Pouls
- The tension is real: AI workloads have long demanded either cloud dependency or bulky hardware, leaving portable professionals caught between convenience and capability.
- ACEMAGIC's F9A disrupts that trade-off by squeezing 192GB of unified memory and 55 TOPS of neural processing into a chassis no larger than a thick hardcover book.
- The unified memory architecture is the critical lever — CPU, GPU, and NPU all draw from a single pool, eliminating the data-shuffling bottlenecks that typically slow down compact machines.
- A live demonstration running DeepSeek V4 Flash, a 284-billion-parameter model, entirely on the device signals that the performance claims are not merely theoretical.
- Two configurations — the PRO 495 flagship and the more accessible Max+ 395 — give the market flexibility, with the device currently landing as a strong statement of where portable AI computing is headed.
At IFA 2026 in Berlin, ACEMAGIC has placed before the world a small but consequential object: a 2-liter workstation housing AMD's new Ryzen AI Max+ PRO 495 processor, capable of running vast artificial intelligence models entirely on-device. It is a quiet challenge to the assumption that serious computation must live in the cloud or in a tower, and a signal that the boundary between portable and powerful is being redrawn. For developers, researchers, and creators who handle sensitive data or simply wish to work untethered, this machine arrives as both a practical tool and a philosophical proposition about where intelligence should reside.
At IFA 2026 in Berlin, ACEMAGIC is demonstrating that serious AI work no longer requires a server room or a cloud subscription. The F9A AI mini workstation, one of the first devices to ship with AMD's new Ryzen AI Max+ PRO 495 processor, fits in roughly 2 liters of space — the footprint of a thick hardcover book — yet carries hardware specifications that would have seemed implausible in a portable machine just a few years ago.
The processor runs 16 cores and 32 threads, boosting to 5.2 GHz, and pairs with integrated Radeon graphics inside what AMD calls a unified memory architecture. Rather than maintaining separate memory pools for the CPU, GPU, and neural processing unit, the system draws from a single shared reservoir — up to 192GB of LPDDR5X in the top configuration. That design eliminates the data-shuffling bottlenecks that typically punish compact machines when handling large models or heavy datasets. The NPU alone delivers 55 trillion operations per second, with total AI throughput reaching 131 TOPS.
ACEMAGIC has demonstrated the F9A running DeepSeek V4 Flash — a 284-billion-parameter language model — entirely on the device, without a cloud connection. Connectivity keeps pace with the ambition: USB4, dual 2.5GbE Ethernet, Wi-Fi 7, Bluetooth 5.4, HDMI 2.1, DisplayPort 2.1, and dual NVMe slots supporting up to 8TB of storage. A second configuration using the Ryzen AI Max+ 395 offers the same core count with slightly lower clock speeds and up to 128GB of memory.
The intended audience is the professional who cannot afford to be tethered — the developer testing inference engines, the researcher protecting sensitive data, the video editor moving between studio and field. With the PRO 495 as its IFA flagship, ACEMAGIC is making a clear argument: powerful, portable, and locally sovereign AI computing is no longer a future category. It is arriving now.
At IFA 2026 in Berlin this week, ACEMAGIC is showing off a machine that fits in your hand but thinks like a workstation. The F9A AI mini workstation, powered by AMD's new Ryzen AI Max+ PRO 495 processor, is one of the first devices to ship with the chip—a processor designed to let people run serious artificial intelligence work locally, without sending data to the cloud.
The machine is small. It measures 158.5 by 158.5 by 81.5 millimeters and occupies about 2 liters of space, the size of a thick hardcover book standing on end. Inside that chassis sits a 16-core, 32-thread processor that can boost to 5.2 gigahertz, paired with integrated Radeon graphics and what AMD calls unified memory—a single pool of RAM that the CPU, GPU, and AI accelerator all draw from. The PRO 495 configuration supports up to 192 gigabytes of LPDDR5X memory, which is the kind of capacity you'd normally expect in a desktop tower, not something you could carry to a coffee shop.
What makes this architecture matter is what it enables. The unified memory means the processor doesn't have to shuffle data between different components—the CPU talks to the GPU talks to the neural processing unit without bottlenecks. ACEMAGIC has demonstrated the F9A running DeepSeek V4 Flash, a 284-billion-parameter language model, entirely on the device. That's the kind of AI work that typically requires either a cloud connection or a much larger machine. The processor delivers up to 55 trillion operations per second of neural processing performance, with total AI performance reaching 131 trillion operations per second, though actual speeds depend on how the software is configured and what task is running.
The connectivity is built for a device meant to work in the field. Two USB4 ports, dual 2.5-gigabit Ethernet connections, Wi-Fi 7, Bluetooth 5.4, HDMI 2.1, and DisplayPort 2.1 give it options for displays and networks. Storage comes from two M.2 NVMe slots that can hold up to 8 terabytes total. ACEMAGIC is also showing a second configuration using the Ryzen AI Max+ 395 processor, which has the same core count and thread count but tops out at 5.1 gigahertz boost and supports up to 128 gigabytes of memory—a step down in capacity but still substantial for a machine this size.
The target market is clear: developers, content creators, and professionals who need to run AI models and memory-heavy workloads without relying on cloud services. A video editor processing footage with AI-assisted tools, a researcher training models on sensitive data, a developer testing inference engines—these are the people ACEMAGIC is building for. The compact form factor means the machine can move between office and home, between lab and field site. The unified memory architecture means it won't choke when handling large datasets or large models. The PRO 495 configuration, with its extra memory and processing headroom, is positioned as the flagship of ACEMAGIC's IFA lineup this year, a statement that the company sees this class of device—powerful, portable, and capable of running AI locally—as where the market is heading.
Citations marquantes
The F9A combines high-capacity unified memory, multi-core CPU performance, integrated graphics, and AI acceleration in a compact workstation intended for developers, content creators, and professional users running AI and memory-intensive workloads locally.— ACEMAGIC product positioning
Envie de l'histoire complète ?
Lire l'original sur manilatimes.net ↗
Technology