AMD has announced the availability of the Ryzen AI Halo developer platform powered by its AI Max 300 series processors. This line of mini PCs won’t win awards for gaming prowess, and they’re not intended as cheap options to plug behind workstation monitors.
For now, AMD will launch just one SKU rocking the Ryzen AI Max+ 395, but later in the year, more models will be released with Ryzen AI Max PRO 400 series chips, some of which will be more affordable. These capable compact boxes will allow developers (and consumers) to run native LLMs with up to 192 GB of unified memory.
What’s in the Halo box?
Everything you need for local models
Powering a single SKU launched in June is strong AMD Ryzen AI Max+ 395 with an impressive 16 physical cores and 32 threads. Capable of up to 5.1GHz, it’s an absolute beast of a mobile processor and has been featured in other products. AMD chose the chip to launch its Ryzen AI Halo developer platform thanks to its inclusion of 80MB of cache, AMD Radeon 8060S graphics, and 650 TOPS NPU. Combine all this with 128GB of built-in storage and you have the ultimate app. runs local LLMs without delay. We love this processor.
With one of these mini PCs, it will be possible to run larger local AI models with all the dedicated RAM. This wild CPU, NPU, and GPU combination streamlines AI development workflows with AMD ROCm support and widely used frameworks and tools. Although AMD is lagging behind Nvidia’s AI push and platform support, the company is definitely trying to make a breakthrough with this thing. 128GB of RAM is perfect for reducing reliance on cloud resources for testing, fine-tuning and development, where even the incredibly popular RTX 3090 would struggle.
The first Ryzen AI Halo won’t be cheap, however, with an MSRP of $3,999. You can thank the cost of the parts for that, as well as the niche nature of the device itself. Pre-orders start in June 2026, but it will be the next wave of SKUs that will become a really interesting offering from AMD. This 300-series AI Halo hasn’t been launched yet, and the company has already announced what’s next with its Ryzen AI Max PRO 400-series CPUs. These Zen 5 chips combine RDNA 3.5 graphics with XDNA 2 NPUs and are among the first x86 client chips to run 300B models, delivering impressive results. So, yes, we are talking about super-major LLMs.
Big improvement with Ryzen AI
AMD has its sights set on Nvidia’s DGX Spark
The Max+ 395 is impressive on its own, but the Max+ PRO 495 and the rest of the 400 series take things even further with up to 192GB of RAM. Compare that to the 16GB or more you use with a discrete GPU to run LLMs at home, and it’s clear to see what a league these are in a league of their own. But it’s not just about launching and developing LLMs. These chips are great for design, rendering, simulation and engineering, making Ryzen AI Halo a great choice for enthusiasts and developers. The best part is how AMD focuses on making it easy to operate.
|
Maximum + 395 |
Max+ PRO 495 |
Max PRO 490 |
Max PRO 485 |
|
|---|---|---|---|---|
|
nuclei Topics |
16 32 |
16 32 |
12 24 |
8 16 |
|
CPU Clock |
Up to 5.1 GHz |
Up to 5.1 GHz |
Up to 5.0 GHz |
Up to 5.0 GHz |
|
Cash |
80 MB |
80 MB |
76 MB |
40 MB |
|
GPU |
Radeon 8060S 40 KI |
Radeon 8065S 40 KI |
Radeon 8050S 32 KI |
Radeon 8050S 32 KI |
|
TDP |
45 – 120 W |
45 – 120 W |
45 – 120 W |
45 – 120 W |
|
NPU (WHO) |
50 |
55 |
50 |
50 |
|
RAM |
128 GB |
192 GB |
192 GB |
192 GB |
Choosing one of the above chips will give you a system capable of handling the biggest modern models. AMD provides the Ryzen AI Developer Center, a ready-to-use software stack for Windows and Linux that includes app and model downloads and playbooks to offer managed workflows. ROCm is fully supported out-of-the-box with optimized performance and SOTA model support, taking full advantage of a wider platform of validated tools, frameworks and drivers. It’s a pretty compelling platform, powered by up to 2 TB of PCIe Gen 4.0 storage, 10Gbps networking, and Wi-Fi 7.
Nvidia already has DGX Spark up and running, but it’s limited to Linux, lacks an NPU, and offers poorer performance, according to AMD. We’re looking forward to testing, as even the Apple Mac Mini M4 Pro and its fantastic AI-optimized hardware and software stack struggle to run models larger than the 100B. Something Ryzen AI Halo can do and then some. And with a 150W TDP, you’re looking at a monthly bill of around $16, at a cost of $0.15 per kWh – not bad compared to the cost of AI cloud platforms.
