AMD launches Ryzen AI Max 400 chips with 192GB memory and $3,999 AI workstation

AMD Ryzen AI Max 400 'Gorgon Halo' packs up to 192GB of unified memory -- refreshed APU uses Zen 5 and RDNA 3.5, and can clock up to 5.2 GHz

They'll show up in Ryzen AI Halo boxes "soon," and pre-orders for Ryzen AI Halo with Strix Halo open in June, starting at $3,999. AMD is refreshing its stack of large SoCs, dubbed Ryzen AI Max, with new Gorgon Point chips. Codenamed Gorgon Halo, the Ryzen AI Max 400 range is a minor refresh to the Ryzen AI Max 300 'Strix Halo' chips already available, similar to what we saw with Gorgon Point in laptops earlier this year. Gorgon Halo comes with one significant difference, however, which is space for up to 192GB of unified memory. With Strix Halo, you could pack up to 128GB of unified memory, but AMD is pushing that boundary higher with Gorgon Halo; perhaps at the worst possible time, as global DRAM shortages continue to push prices up across all business categories. It'll be a minor miracle if AMD is able to actually ship Gorgon Point with 192GB of unified memory consistently -- we've seen Apple remove the 512GB option and even the 128GB option from the Mac Studio due to memory shortages. Regardless, AMD has a lineup of three chips that should look very familiar if you've looked over the Strix Halo stack. All three chips use Zen 5 CPU cores and RDNA 3.5 GPU cores, alongside an XDNA 2 NPU. The flagship Ryzen AI Max+ Pro 495 comes with a minor clock speed bump of 100 MHz over the Ryzen AI Max+ 395, allowing it to boost to 5.2 GHz. Otherwise, you could scratch off the "4" and replace it with a "3," at least based on the specs AMD has shared so far. These chips currently have a "Pro" tag, which means they're targeting the commercial market. However, the slides below refer to the Ryzen AI Max 400 range more broadly. I asked AMD about this discrepancy, and a spokesperson sent the following: "AMD will be announcing the Ryzen AI Max PRO 400 Series, featuring AMD PRO technologies which deliver enterprise-grade security, manageability, and reliability." So, I guess consumer Gorgon Halo is still up in the air. Maybe. Interestingly, AMD opted to stick with a GPU with 32 CUs (the Radeon 8050S) for the Ryzen AI Max Pro 490 and 485. Earlier this year, AMD refreshed the Ryzen AI Max 385 and 390 with 40 CUs, the same as the flagship, in the form of the Ryzen AI Max+ 388 and 392. Maybe we'll see a refresh of the refreshed Gorgon Halo chips with 40 CUs down the line. Memory is the big upgrade here. Regardless of the GPU configuration, up to 160GB of unified memory can function as VRAM (32GB is reserved for the system). AMD says that much memory makes Ryzen AI Max 400 chips the first x86 client processors able to run a 300B+ parameter LLM. It wins in a category of one, however: Intel doesn't make a large SoC like Gorgon Halo, and Apple uses the ARM ISA. AMD says Ryzen AI Max 400 chips are "coming soon," but didn't share any timeline beyond that, nor any partners for Gorgon Halo systems. Strix Halo, as a niche product, was only available in a handful of machines, such as the Framework Desktop, ROG Flow Z13, and GMKtec EVO-X2. There's a good chance we'll see a similarly conservative rollout of Gorgon Halo, as well. Despite not sharing any partners, AMD tells me that "several OEM partners have expressed excitement for the Ryzen AI Halo platform and the Ryzen AI Max Pro 400 series family of processors," and that "systems will be announced from our partners starting in Q3 2026." AMD Ryzen AI Halo starts at $3,999 -- pre-orders in June The only confirmed machine with Ryzen AI Max 400 so far is the Ryzen AI Halo, which is "coming soon" configured with the Ryzen AI Max+ Pro 495. Coming sooner is the Ryzen AI Halo box AMD revealed earlier this year with the Ryzen AI Max+ 395. AMD is opening up pre-orders in June, and it says the machine starts at $3,999. The starting configuration includes a Ryzen AI Max+ 395 with 128GB of unified memory and 2TB of storage. It says "details and pricing of the other configurations will be released closer to on-shelf," so it sounds like we'll see additional models in the future. The Ryzen AI Halo's main competitor, Nvidia's DGX Spark, is currently selling for $4,700 with 128GB of unified memory, Nvidia's GB10 chip, and 4TB of storage. The Ryzen AI Halo supports Linux and Windows, while the DGX Spark is limited to Linux. Still, the penguin seems like the primary platform for this box. With Linux, AMD says the Ryzen AI Halo offers up to 14% higher tokens per second than the DGX Spark with the GLM 4.7 Flash 30B model, as well as up to 4% higher tokens per second with Qwen 3.6 35B. AMD also compared the Ryzen AI Halo to the Mac Mini M4 Pro, showing around 4X scaling in AI workloads. That's not really a fair comparison, however; a Mac Studio is more akin to the level of compute inside the Ryzen AI Halo or DGX Spark. Outside of the core components, the Ryzen AI Halo comes with Wi-Fi 7, Bluetooth 5.4, and 10Gbps Ethernet, along with an HDMI 2.1b display output. The device includes three USB-C ports (no word on speeds yet), along with a fourth USB-C used for power delivery. The rated TDP is up to 120W for the box. It's a lot of hardware crammed into a tiny space, with the Ryzen AI Halo coming in at 5.9 x 5.9 x 1.7 inches. Although these boxes are expensive, AMD is framing the Ryzen AI Halo around the "token economy," similar to how Nvidia has messaged against its data center hardware. AMD says one Ryzen AI Halo box can save up to $750 each month over using cloud compute, claiming the Ryzen AI Halo will break even on cost after six months (assuming six million tokens per day). With AI agents, that token usage is certainly possible. Just this month, we saw OpenClaw developer Peter Steinberger rack up $1.3 million in OpenAI API usage in just 30 days across a three-person team working on the agentic AI framework. Pre-orders open in June for the Ryzen AI Halo., although AMD hasn't shared the exact date. For the updated Ryzen AI Halo box with Gorgon Halo chips, AMD hasn't announced any release date yet. Assuming we see more systems in Q3 as AMD has suggested, we should have a better idea about the Gorgon Halo rollout at that point. Full presentation Follow Tom's Hardware on Google News, or add us as a preferred source, to get our latest news, analysis, & reviews in your feeds.

AMD says its $4K Ryzen AI Halo workstation practically pays for itself

AMD's answer to Nvidia's DGX Spark AI workstations, codenamed the Ryzen AI Halo, will be available for pre-order later next month for anyone with $3,999 burning a hole in their pocket. That might sound like a lot for an AI mini PC, but don't worry. Compared to cloud APIs, it practically pays for itself. Or, well, that's AMD's sales pitch. The House of Zen argues that if you spend eight hours a day vibe coding, the system could save you $750 a month. Whether this helps you justify paying for hardware that less than a year ago could be found for between $2,200 and $2,999 or not, it's (probably) not AMD being greedy here; the RAMpocalypse has been hard on everyone. Much like the DGX Spark, which now retails for $4,699, up from $3,999 when we reviewed it last fall, AMD's rendition aims to provide a curated developer environment for running local models and agentic AI frameworks. This is really the core value proposition behind either of these devices. They aren't the most powerful or fastest AI systems, but they're able to run models that a few years ago would have cost $20K or more. The diminutive system measures in at 5.9 x 5.9 x 1.7 inches (150 x 150 x 43 mm) and is powered by a 120 watt Ryzen AI Max+ 395 APU, better known by its codename Strix Halo. The chip is backed by 128 GB of LPDDR5x 8000 MT/s memory, which feeds both its 16 Zen 5 cores and 40 RDNA 3.5 GPU compute units, providing up to 256 GB/s of bandwidth, more than a Ryzen 9000 Threadripper (non-Pro) system. For local AI enthusiasts, that's enough to run models up to 200 billion parameters in size at 4-bit precision -- just like the more expensive Spark. The bulk of the Ryzen AI Halo's compute comes from its integrated graphics, which are capable of delivering roughly 56 teraFLOPS at 16-bit precision. While impressive for onboard graphics, that's still between 55 and 88 percent slower than what the DGX Spark advertises. Unlike the Spark's Blackwell-based GB10 APU, Strix Halo doesn't support FP8 or FP4 data types in hardware. At BF16, the Spark delivers 125, at FP8 250, and FP4 500 teraFLOPS. Double those figures if you happen to find a workload that can leverage Nvidia's 4:2 sparsity. That performance discrepancy won't necessarily be obvious in every workload. In fact, in LLM inference, AMD claims the AI Halo generates tokens 4-14 percent faster than the Spark. The lower end of that roughly matches what we saw when we pitted the Spark against a similarly equipped HP Z2 Mini G1a back in December. The G1a packs the same silicon as AI Halo, and in Llama.cpp with the Vulkan backend, eked out a small but meaningful lead over the Spark in tokens per second generated. However, the speed any GPU can generate tokens at is largely dictated by effective memory bandwidth, not floating point performance. GPU compute has a much bigger impact on things like prompt processing time. In our testing, the Spark's more capable tensor cores gave it a 2-3x lead in prompt processing. For shorter prompts, this isn't all that noticeable, usually the difference between waiting 100 ms versus 200 ms or 300 ms, but for longer prompts, it did become more pronounced. We saw the Spark take similar leads in our image generation and fine tuning benchmarks, but it's worth noting that AMD's software stack has matured greatly since our initial review and the performance gap has likely closed somewhat since then. AMD's AI Halo does have two things going for it that can't be said of the Spark. Alongside the GPU is an XDNA 2-based neural processing unit (NPU) that AMD rates for 50 TOPS. What good that'll do you depends heavily on the application in question. Many content creation apps have now been updated to take advantage of it, but the number of generative AI inference engines that could properly harness it was quite limited the last time we looked. The second thing AMD's Ryzen AI Halo has going for it is that it's a standard x86 box at its heart, and you can run Windows or your preferred flavor of Linux on it if that's more your style. On the Spark, you're stuck with a lightly customized version of Ubuntu 24.04. Beyond that, you're coloring outside the lines. Particularly for developers building for Microsoft's NPU-accelerated AI PC ecosystem, this is an obvious advantage. In terms of networking, AMD's Spark-clone falls a bit flat. One of the hallmark features of Nvidia's AI workstation is a 200 Gbps ConnectX-7 NIC, which allows for clustering of up to two and eventually four systems. AMD's AI Halo has a single 10 Gbps NIC, which should help with downloading large model files in a timely manner. In theory, the system should be able to achieve high-speed networking over USB-4, but it's not clear whether this is actually a supported use case. That said, Apple has already demonstrated just this using RDMA over Thunderbolt, so it should work so long as AMD has a playbook for configuring RDMA on its systems. As we mentioned earlier, much of the Ryzen AI Halo's value proposition comes from being validated hardware with well documented playbooks for common use cases and known good software. Finding the right combination of device drivers, ROCm, HIP, SYCL, CUDA, PyTorch, TensorFlow, and JAX has long plagued the AI/ML devs regardless of which ecosystem you opt for. Having validated environments for workloads, whether it be vLLM, Llama.cpp, Ollama, ComfyUI, or something else ensures users spend more time doing something productive than debugging mismatched dependencies. At launch, AMD says the Ryzen AI Halo will ship with five preinstalled playbooks, with another 10 available online and additional playbooks to be added monthly. Additionally, customers will gain access to AMD's developer program, cloud credits, and exclusive playbooks. The 128 GB Ryzen AI Halo will be available for pre-order next month starting at $3,999, but if that isn't enough for you, AMD is already prepping a higher capacity version of the system with 192 GB of memory on board. That system will feature a refreshed Ryzen APU in the AI Max+ 495, which just like the rest of AMD's 400-series lineup gets a modest clock bump to the CPU, GPU, and NPU, and not a whole lot else. Still, 192 GB of unified memory opens the door to even larger, more capable models, if you can stomach the presumably higher asking price. ®

AMD just dropped a compact AI workstation that makes discrete GPUs look outdated for running LLMs

Richard is the PC Hardware Lead at XDA and has been covering the technology industry for almost two decades. He's been building PCs since young, and when not creating content, you can often find him inside a chassis somewhere. AMD has announced the availability of the Ryzen AI Halo developer platform, powered by AI Max 300-series processors. This range of mini PCs isn't going to win awards for gaming prowess, nor are they designed as low-cost options for attaching behind workstation monitors. For the time being, AMD will offer just one SKU for sale, rocking the Ryzen AI Max+ 395, but more models will roll out later in the year with Ryzen AI Max PRO 400-series chips, some of which will likely be more affordable. These capable compact boxes will allow developers (and prosumers) to run local LLMs with up to 192 GB of unified memory. What's in the Halo box? Everything you need for local models Powering the single SKU launching in June is the mighty AMD Ryzen AI Max+ 395 with its impressive 16 physical cores and 32 threads. Capable of boosting up to 5.1 GHz, this is an absolute beast of a mobile processor and has been featured in other products. AMD chose the chip to launch the Ryzen AI Halo developer platform thanks to the inclusion of 80 MB of cache, AMD Radeon 8060S graphics, and a 650 TOPS NPU. Combine all of that with 128 GB of unified memory, and you've got the ultimate application for running local LLMs without delay. We love this processor. With one of these mini PCs, it'll be possible to run larger local AI models with all that dedicated RAM. That beastly CPU, NPU, and GPU combo ensures that AI development workflows are streamlined, with support for AMD ROCm and widely used frameworks and tools. Though AMD has lagged behind Nvidia's AI push and platform support, the company is certainly looking to cause a splash with this thing. 128GB of RAM is perfect for reducing reliance on cloud resources for testing, fine-tuning, and development, something that even the incredibly popular RTX 3090 would struggle with. The first Ryzen AI Halo won't be cheap, however, with an MSRP of $3,999. You can thank the pricing of parts for that, as well as the niche nature of the device itself. Pre-orders commence in June 2026, but it'll be the next wave of SKUs that will really make this an interesting proposition by AMD. This 300-series AI Halo hasn't even launched yet, and the company has already outed what's next with Ryzen AI Max PRO 400 series CPUs. These Zen 5 chips combine RDNA 3.5 graphics with XDNA 2 NPUs to deliver some impressive results, being among the first x86 client chips to run 300B models. So, yes, we're talking super-large LLMs. I finally found a local LLM I want to use every day (and it's not for coding) Local AI that actually fits into my day Posts 19 By Nolen Jonker Going big with Ryzen AI AMD has its sights 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 up a notch further with up to 192 GB of RAM. Compare that to the 16 GB or so you're using with a discrete GPU for running LLMs at home, and it's clear to see what league these are in a league of their own. But it's not just for running and developing LLMs. These chips are great for design, rendering, simulation, and engineering, making the Ryzen AI Halo quite the option for enthusiasts and developers. The best part is how AMD focused on making it easier to get up and running. Max+ 395 Max+ PRO 495 Max PRO 490 Max PRO 485 Cores Threads 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 Cache 80 MB 80 MB 76 MB 40 MB GPU Radeon 8060S 40 CUs Radeon 8065S 40 CUs Radeon 8050S 32 CUs Radeon 8050S 32 CUs TDP 45 - 120 W 45 - 120 W 45 - 120 W 45 - 120 W NPU (TOPS) 50 55 50 50 RAM 128 GB 192 GB 192 GB 192 GB Picking one with the above chips will land you a system capable of handling the largest modern models. AMD provides a ready-to-use software stack for Windows and Linux, consisting of the Ryzen AI Developer Center, apps and model preloads, and playbooks to offer guided workflows. ROCm is fully supported out of the box with optimized performance and SOTA model support, fully utilizing the wider platform of validated tools, frameworks, and drivers. It's quite the compelling platform, bolstered with up to 2 TB of PCIe Gen 4.0 storage, 10Gbps networking, and Wi-Fi 7. Get Ryzen AI Halo Hardware Insights -- Subscribe Subscribe to the newsletter for hands-on analysis of AMD's Ryzen AI Halo platform, practical guidance for running local LLMs, ROCm and tooling breakdowns, and clear model-sizing insights - all as part of our AI hardware coverage. Get Updates By subscribing, you agree to receive newsletter and marketing emails, and accept our Terms of Use and Privacy Policy. You can unsubscribe anytime. Nvidia already has the DGX Spark out and running, but it's limited to Linux, lacks an NPU, and offers weaker performance, according to AMD's data. We look forward to getting our hands on one to test, as even the Apple Mac Mini M4 Pro and its fantastic optimized hardware and software stack for running AI, struggles to run models larger than 100B, which is something the Ryzen AI Halo can do and then some. And with a TDP of 150W, you're looking at a monthly bill of around $16 with a per kWh price of $0.15 -- Not bad when compared to the price of AI cloud platforms.

AMD prices its Ryzen AI Halo PC at $3,999, unveils Ryzen AI Max 400 chips - Engadget

AMD's big pitch for 2026 seems to be: "Who needs cloud AI processing when you can do it all locally?" At CES this year, the company unveiled its Ryzen AI Halo PC, a Mac Mini-sized system that can crank out AI work. Today, AMD announced that it will start at $3,999 with Ryzen AI Max 300 CPUs, and we can also look forward to a future model with new Ryzen AI Max 400 chips, as well. Preorders start in June. While pricey, AMD positions the Halo as a cost-effective alternative to paying high monthly AI computing fees. If you're spending $773 a month to use 6 million daily AI tokens -- which isn't an unusual scenario for many developers -- the Halo could pay itself off within six months. And for more demanding work, AMD says its $4,000 Radeon R9700 Pro GPU could break even within three months for people paying $2,253 a month to use 18 million daily tokens. If it's not clear already, these aren't devices meant for regular consumers. Instead, AMD is directly competing with NVIDIA's DGX Spark AI PC, which now goes for $4,699 after launching at $4,000. While NVIDIA's AI PC can only run Linux, the Ryzen AI Halo can run either Windows or Linux, since it's powered by an x64 chip. Another advantage? The Halo has a 50 TOPS NPU and a Radeon GPU with 40 compute units, whereas the DGX Spark leans entirely on NVIDIA's Blackwell GPU for AI work. Both systems also have 128GB of unified system memory, which is essential for running large models. Notably, that's also more memory than you can have in a Mac Mini or Mac Studio, which are both popular with AI developers. As for those new Ryzen AI Max 400 chips, they'll be led by the AI Max+ Pro 495, a 16-core chip with a 5.2GHz boost speed, 55 TOPS NPU and Radeon 8065S graphics. Those chips will also support up to 192GB of unified memory, allowing for 160GB of GPU VRAM. Spec-wise, it's only slightly faster than the AI Max 395, which has a 5GHz CPU boost clock speed, but we've yet to see comparison benchmarks from AMD. The company says Ryzen AI Max 400 chips will be available in the third quarter of 2026.

AMD’s Next Big Chip Hopes To Beat Nvidia’s CPUs While They’re in the Crib

Team Red also hopes to beat Nvidia’s â€~AI supercomputer’ with a mini-PC of its own. AMD is trying to give Nvidia’s long-rumored laptop CPUs a run for their money before they’re even out the door. To do this, AMD is making its most powerful processor with the latest Ryzen AI Halo chips thatâ€"hopefullyâ€"won't be a no-show like its recent Ryzen AI 400 laptop CPUs. The newly revealed Ryzen AI Max+ Pro 495 chip seems like a beast, at least on paper. The main draw is the newer Zen 5 CPU, which features 16 cores and 32 threads with a boost clock speed of 5.2GHz. However, the new chip has the same old RDNA 3.5 GPU microarchitecture that's inside the previous-generation Ryzen AI Max+ 395. That Radeon 8065S graphics chip includes 40 compute units (the name AMD gives to its core clusters). It should also get a small boost with up to 160GB of possible VRAM (video random access memory), pushing the GPU memory further than before. Without AMD’s latest RDNA 4 GPU microarchitecture, it's hard to call it a huge jump in possible performance compared to the Ryzen Max+ 300 series. AMD claims the Ryzen AI Max+ Pro 495 is its first x86 processor that can run a 300 billion parameter AI model all by itself. AMD also promoted two other versions of the same APU (accelerated processing unit, which combines CPU and GPU capabilities). There’s a Ryzen AI Max Pro 490 with 12 cores and 24 threads, plus a Ryzen AI Max Pro 485 with eight cores and 16 threads. The lower-spec models use a smaller GPU with only 32 compute units. The reason why AMD’s last generation of Strix Halo chips became so popular wasn't necessarily because of AI capabilities but rather because of their surprisingly capable graphics performance. This year, AMD expanded the lineup with a few toned-down Strix Halo processors specifically for gaming and creative-centric devices. The side effect of the AI-induced RAM crisis is that several companies cancelled gaming handhelds that were supposed to feature those toned-down processors. As such, while we’re losing out on boutique gaming products, AMD decided it was time to launch a miniature “AI developer platform†called AMD Ryzen AI Halo. The small PC contains AMD’s own Ryzen AI Max+ 395, the previous-gen Strix Halo chip and not its newfangled Gorgon Halo design. It otherwise supports a 2TB SSD and 128GB of unified memory. We’ve seen this chip perform well for mobile and pint-sized devices like the Asus ProArt PX13 GoPro Edition and the Framework Desktop. The announcements were part of a wave of AI-centric news as AMD desperately tries to claw away an ounce of Nvidia’s clout with AI developers for itself. Nvidia has already released its $4,000 DGX Spark, a desktop AI-centric computer built on an ARM chip and the company’s Blackwell GPU architecture. AMD’s Ryzen AI Halo costs $4,000 for a box that measures just 6 by 6 inches. The DGX Spark also runs on a version of Linux, whereas AMD’s Ryzen AI Halo can use Linux as well as Windows for a more comfortable PC-like experience. Running on an x86 chip and Windows will effectively mean the AI Halo will work just as well as any other mini PC for regular day-to-day tasks and not just natively running OpenClaw. Apple’s latest Mac mini has proved especially popular for running on-device AI, though AMD promises its AI Max+ chip offers 4X generative AI workloads compared to an M4 Pro. The Ryzen AI Halo with the previous-gen 395 chip should be available for preorder in June, according to the company. AMD added that a version of the device with Ryzen AI Max+ Pro 495 is “coming soon.†As far as the new 495+ chip goes, the company promised we’ll have new devices from Asus, HP, and Lenovo sometime in Q3 this year. Nvidia’s upcoming laptop CPUs, which are likely to hit the scene at Computex in June, will now have to compete against AMD's latest and Intel’s Panther Lake chips for laptop dominance.

This $3,999 AMD mini PC replaces expensive cloud AI without the Nvidia price tag

AMD might have the solution if you like the idea of Nvidia's DGX Spark as an AI workstation, but balk at having to use a specialized ARM chip -- and the $4,699 starting price. The company has introduced the Ryzen AI Halo, a mini PC that's not only optimized for AI development, but promises to save money both up front and by avoiding costly subscriptions. The new system is built around a Ryzen AI Max CPU, whether it's the longstanding Max+ 395 (Strix Halo) or a new Max+ Pro 495 (Gorgon Halo). The use of 16 Zen 5 CPU cores, unified memory, and as much as a 40-core integrated GPU lets the Ryzen AI Halo run many large AI models locally without choking, or even consuming much space -- despite a 5.9in by 5.9in footprint, the base model fits up to 128GB of RAM and a 2TB SSD. AMD claims up to 50 TOPS of AI processing from the built-in NPU alone. Framework Desktop Brand Framework CPU AMD Ryzen AI Max 300-series $1099 at Framework Expand Collapse The Ryzen AI Max+ Pro 495 doesn't offer much more raw computing power with the same core count, GPU capabilities, and 55-TOPS NPU. However, it supports up to 192GB of RAM that could be vital for larger projects. AMD claims the Ryzen AI Halo has some raw performance advantages over the DGX Spark. The gains in tokens per second range from 4 percent for Qwen 3.6 through to 14 percent for GLM 4.7 Flash. It also notes that an M4 Pro-based Mac mini tops out at 64GB of RAM, so you can't run local versions of Qwen 3.5 or GPT OSS. AMD's Ryzen AI Halo software edge: It runs Windows, too You can run Linux, but it's not mandatory AMD believes Ryzen AI Halo software options also make it a better choice over Nvidia's AI computer. It supports both Windows and Linux, so you can use Windows tools if you need them. This may be a better choice than the DGX Spark if you want your development box to double as an ordinary PC. You'll also get preloaded apps and models with "Playbooks" to guide developers new to these tools. Out of the box, you can expect optimized models like GPT-OSS, FLUX 2, and SDXL. There's also support for "leading" AI models, AMD says. A Ryzen AI Developer Center both syncs software across devices and lets you update or revert apps from a central hub. AMD Ryzen AI Halo price and availability A potential bargain if you hate subscriptions The Ryzen AI Halo with the Max+ 395 CPU will be available for pre-order in June starting at $3,999, with the Max+ Pro 495 version as yet unpriced and "coming soon." That's a significant discount over the $4,699 DGX Spark, although you don't get Nvidia's 4TB of storage. Related Trying local AI models became way easier after I installed this app No fussing with the command line. Posts By Nick Lewis With that said, AMD is betting that you'll save money if you previously depended on cloud computing for your AI work. If you use the Claude Sonnet 4.5 developer framework, you'll theoretically save up to $750 per month if you use it for eight hours per day. The savings climb to $2,200 per month if you're relying on a dedicated GPU like AMD's own Radeon AI Pro R9700. Your actual savings (if any) will depend on the models you use and what you're trying to accomplish. A more conventional desktop with a faster GPU and more memory will still be better for the most demanding users. Like Nvidia, however, AMD is more focused on efficiency, and on giving you a workstation that can sit alongside a conventional PC. If you only occasionally need all-out AI processing, the Ryzen AI Halo might be a better value simply because it's easier to use as your only PC, particularly if you need Windows.

AMD Prices Ryzen AI Halo Mini PC at $3,999

AMD has officially disclosed pricing for its Ryzen AI Halo mini PC platform, confirming that the compact AI-focused system will launch at $3,999 with pre-orders scheduled for later in June. The announcement also introduces the Ryzen AI Max 400 processor family, internally codenamed "Gorgon Halo," which refreshes the earlier Ryzen AI Max 300 series with updated clocks, higher AI throughput, and expanded unified memory support. The Ryzen AI Halo mini PC is built around the Ryzen AI Max+ 395 processor and targets developers running local AI inference workloads, large language models, and workstation-class accelerated applications. AMD is positioning the system as an alternative to cloud-heavy development environments, claiming that local deployment on the hardware could reduce cloud compute expenses by roughly $750 per month depending on workload usage. Physically, the system uses a compact 5.9 x 5.9 x 1.7-inch chassis, but AMD equips it with specifications typically associated with much larger workstation hardware. The platform ships with 128GB of LPDDR5X memory alongside a 2TB PCIe Gen4 x4 SSD. Unified memory capacity is a major focus for the Halo platform because larger AI models increasingly depend on high-capacity local memory pools instead of traditional discrete GPU memory limits. Connectivity includes a 10 Gigabit Ethernet port, Wi-Fi 7, Bluetooth 5.4, HDMI 2.1b, three USB-C ports, and a dedicated USB-C power delivery connection. AMD confirmed support for both Windows and Linux operating systems, broadening compatibility for enterprise and development workflows. Alongside the mini PC announcement, AMD detailed the Ryzen AI Max 400 series itself. Compared to the Ryzen AI Max 300 lineup, the refresh brings relatively modest CPU and GPU frequency improvements of around 0.1GHz, but increases NPU AI compute capability by an additional 5 TOPS. More importantly, the platform now supports up to 192GB of unified memory, with as much as 160GB configurable as shared graphics memory. That expanded memory capacity may become increasingly relevant as local AI inference workloads continue to scale. Running larger models directly on client hardware remains heavily constrained by memory allocation and bandwidth, making unified memory one of the more important platform characteristics for compact AI systems. The Ryzen AI Halo launch further signals AMD's growing interest in AI-focused client hardware that blends CPU, GPU, and NPU acceleration into compact workstation-class designs.

AMD's $3999 Halo box challenges NVIDIA's DGX Spark with 128GB memory and Zen 5 AI power

AMD says its Ryzen AI Halo mini-PC will "pay for itself" in just a few months of use and it's coming in June. The compact system, built around the Ryzen AI MAX+ 395 chip and packing 128GB of unified memory, is AMD's direct answer to NVIDIA's DGX Spark that debuted in October 2025. Priced at $3999, the Halo Box is aimed squarely at local AI development and enterprise workloads that demand high performance and portability. The Ryzen AI Halo is based on AMD's Strix Halo platform and features a powerful integrated RDNA 3.5 GPU, 16 Zen 5 cores, and support for both Windows and Linux. It's designed to run complex AI models locally without relying on cloud infrastructure, making it ideal for developers and businesses focused on privacy and latency-sensitive applications. AMD claims the system will deliver enough efficiency and performance to offset its cost quickly in real-world use cases. This move signals AMD's growing push into the AI workstation and AI-focused mini-PC market. By offering a self-contained, high-performance solution at a competitive price, the company is attempting to compete in a space currently dominated by NVIDIA. The Halo Box is also one of the many key steps in AMD's broad strategy to expand its presence in various AI markets. With a June launch window confirmed, AMD is now focused on delivering the Halo Box as a viable alternative to more expensive, cloud-dependent solutions. If the system can live up to its promises, it could shift the landscape for local AI development and give developers a more affordable, portable option for running large models.

AMD Claims Leadership Tokens/$ With Its Ryzen AI Halo Dev Platform, Tackles NVIDIA's Spark at $3999 & Pays For Itself Within 6-Months

AMD's Ryzen AI Halo developer platform offers strong Agentic AI capabilities & "leadership" token/$ value with its Halo chips, and will be available for pre-order this June. Today, AMD is finally lifting the pricing and availability details of its Ryzen AI Halo developer platform, which was first announced at CES 2026, followed by a recent tease at its AI Dev Day. But before that, let's talk a little bit about the platform itself and its major highlights, which are listed below: The AMD Ryzen AI Halo developer platform launching this June will be based on the Ryzen AI MAX 300 family, codenamed Strix Halo, which has seen some major adoption in the past few months, from laptops to handhelds and Mini PCs; it's entering every consumer PC segment. These high-performance and premium SoCs offer amazing performance thanks to their Zen 5 CPU, RDNA 3.5 GPU, and XDNA 2 NPU architectures. The Ryzen AI Halo combines these SoCs in a small form factor for developers and SFF AI users. AMD's Ryzen AI Halo is based on the flagship Strix Halo SoC, the Ryzen AI MAX+ 395, which features 16 cores, 32 threads based on the Zen 5 architecture, the Radeon 8060S iGPU with 40 RDNA 3.5 cores, a 50 TOPS XDNA 2 NPU, and a TDP of up to 120W. The PC will be equipped with 128 GB LPDDR5X-8000 RAM and 2 TB of PCIe Gen4x4 storage. In addition to that, the platform itself measures just 5.9" x 5.9" x 1.7", making it ultra-compact and shorter than Apple's Mac Mini Pro (M4). It comes with 3 USB Type-C ports, including one for power input, Wi-Fi 7, BT 5.4, 10 Gbps Ethernet, and HDMI 2.1b. The Ryzen AI Halo Mini PC will feature full AMD ROCm Support, including the newly released ROCm 7.2.2 suite, will be optimized for Dev-Ready applications such as LM Studio, ComfyUI, VS Code, and More, will enable optimizations for several models, including GPT-OSS, FLUX.2, SDXL, and More, and it will carry Day 0 support for leading AI models. AMD compares the Ryzen AI Halo Mini PC against two competitors: DGX Spark from NVIDIA and Mac Mini M4 Pro from Apple. Versus the DGX Spark, AMD claims the Ryzen AI Halo offers wider OS support, leadership LLM Token value, and includes an NPU rated at 50 TOPS. The company also showcases some AI numbers against DGX Spark, which are listed below: Compared to the Apple Mac Mini, the Ryzen AI Halo offers twice the max memory config as the M4 Pro, can run up to 200B models, whereas the Mac Mini can't go beyond 100B models, and offers broader Gen AI capabilities. AMD states that the AI Halo is on average 4x faster than the M4 Mac Mini Pro. One of the biggest advantages highlighted by AMD for its Ryzen AI Halo is its higher token value and its ability to pay for itself. AMD states that not every agent and workflow needs a frontier model, and most of the grunt work can be shifted to local instead of the cloud. Developers running localized AI can save up to $750 USD per month when switching from Cloud-based AI. For example, with AMD Ryzen AI Halo, you will pay the initial $3999 price, followed by a $16.2 monthly electricity cost, which is measured at a sustained 150W draw (cited as a nightmare case). Meanwhile, the AI Cloud services net you roughly $750 per month, assuming up to 31 Million Tokens (8 hours/day) at 36 Tokens/s or up to 385 Million Tokens (8 hours/day) assuming 446 Tokens/s. Given this math, AMD's Ryzen AI Halo will be able to Break-Even in just 6 months, and the total bill after 3 years will be roughly around $4500-$4600 versus the cloud services for which you'll be paying over $25K. Following the pre-orders of the Ryzen AI Halo (Ryzen AI MAX+ 395) in June 2026 for $3999, AMD is also going to introduce an updated variant with its Ryzen AI MAX+ 495 SoCs around Q3 2026, which will come packed with even more capabilities and 192 GB of memory, enabling 300B+ model support. As for the price itself, it is competitive against the NVIDIA DGX Spark, which costs $4679 right now, but at the same time, there are several Ryzen AI MAX+ Mini PCs with similar configurations that come in at a lower price. Regardless, the AMD Ryzen AI MAX SoCs are superb, and we have our own review here that proves this.