Beelink GTR9 Pro AMD Ryzen™ AI Max+ 395 Processor
Original price was: $4,699.00.$4,349.00Current price is: $4,349.00.
16 Zen 5 CPU cores, combined with the advanced Radeon™ 8060S iGPU, next-gen XDNA 2 NPU, and 126 AI TOPS, deliver cutting-edge architecture that significantly boosts the GTR9 Pro’s performance.
The Beelink GTR9 Pro is built for one job most mini PCs can’t do: running large local AI models entirely on-device, at real speed, with room to spare. Powered by AMD’s flagship Ryzen AI Max+ 395 — 16 Zen 5 cores, a Radeon 8060S iGPU, and an XDNA 2 NPU delivering 126 AI TOPS — paired with a huge 128GB of unified LPDDR5X-8000 memory, it’s aimed at developers, researchers, and homelabbers who want data-centre-class AI capability on their own desk, without cloud costs or data leaving the building.
Key Features
- AMD Ryzen™ AI Max+ 395 CPU and Radeon™ 8060S GPU — 16 Zen 5 CPU cores combined with the advanced Radeon 8060S iGPU, a next-gen XDNA 2 NPU, and 126 AI TOPS deliver cutting-edge architecture that significantly boosts performance for AI, creative, and everyday workloads.
- 128GB Unified LPDDR5X-8000 Memory — shared between CPU, GPU, and NPU, this is enough headroom to load large quantized language models like DeepSeek 70B entirely in memory for private, local inference — no API costs, no data leaving your network.
- 140W Ultra-Quiet Cooling — dual turbine fans plus a full-coverage vapor chamber keep the chip running at 140W TDP at just 32dB, delivering massive sustained performance with near-silent operation even under full load.
- Dual M.2 2280 PCIe 4.0 Storage — two NVMe slots support up to 16TB combined (8TB per slot); the retail unit ships with a 2TB SSD rated up to 7000MB/s.
- AI Server Clustering — dual 10GbE LAN ports and dual USB4 (40Gbps) let the GTR9 Pro act as a local AI computing hub, whether that’s serving models to other machines on your network or linking multiple units for larger workloads.
- Quad 8K Display Support — HDMI 2.1, DisplayPort 2.1, and dual USB4 ports (40Gbps, 8K@60Hz) drive up to four 8K displays, ideal for data-dense workspaces and multi-monitor development setups.
- Industrial-Grade Build — an all-metal chassis with an internal aluminium frame and a built-in 230W PSU is engineered for long-term stability under sustained AI workloads, not just short bursts.
- AI Voice Pickup and Dual Speakers — a built-in microphone with AI-powered 360° pickup separates voice from background noise within a 5-metre range, while DSP-tuned dual speakers handle calls and media without external equipment.
Built for Local AI, Not Just Everyday Computing
The reason buyers choose the GTR9 Pro over a conventional mini PC comes down to memory: 128GB shared across CPU, GPU, and NPU means it can hold large open-weight models entirely in RAM and run inference locally, with no per-token API costs and no dependency on a third party’s uptime. For developers experimenting with local LLMs, researchers fine-tuning smaller models, or small teams wanting to keep sensitive data in-house, that’s the whole point of this machine — and the dual 10GbE networking means it can just as easily serve as a shared inference hub for a small homelab or office rather than a single-user workstation.
It’s just as capable outside AI workloads. The Radeon 8060S iGPU and 16 Zen 5 cores handle everyday multitasking, light creative work, and even casual gaming without breaking a sweat, while the quad 8K display support and dual USB4 ports make it a genuinely capable desktop replacement for anyone who needs a lot of screen real estate. The near-silent 32dB cooling means it’s comfortable to run on a desk all day, not tucked away in a server closet.
Every unit ships from our UK, EU, or US warehouse depending on your delivery address, with the correct regional power adapter included, and the power supply itself accepts anywhere from 100V to 240V regardless of what the label states — so it’s ready to use wherever you are without a separate voltage converter.
Specifications
- Processor: AMD Ryzen AI Max+ 395 (16-core, Zen 5)
- Graphics: AMD Radeon 8060S iGPU
- AI acceleration: XDNA 2 NPU, up to 126 AI TOPS combined
- Memory: 128GB unified LPDDR5X-8000
- Storage: 2x M.2 2280 PCIe 4.0 slots (up to 8TB each, 16TB combined); 2TB SSD included, rated up to 7000MB/s
- Networking: 2x 10GbE LAN, Wi-Fi
- Ports: 2x USB4 (40Gbps), HDMI 2.1, DisplayPort 2.1
- Display support: up to four 8K displays
- Cooling: dual turbine fans, full-coverage vapor chamber, 140W TDP at ~32dB
- Power supply: built-in 230W PSU, 100–240V wide-range input
- Chassis: all-metal housing with internal aluminium frame
- Audio: AI-enhanced 360° microphone pickup (5m range), dual DSP-tuned speakers
| RAM+SSD | 128GB LPDDR5x-8000 memory (soldered) + 2TB SSD |
|---|---|
| Color | Frost Silver |
| Power Adapter | AU, EU(Pre order), JP, UK, US |
14 reviews for Beelink GTR9 Pro AMD Ryzen™ AI Max+ 395 Processor
Fast, Tracked Worldwide Shipping
In-stock orders are processed within 1–3 business days and shipped with full tracking. Delivery costs and timeframes depend on your destination:
Delivery Estimates & Costs
- United Kingdom — 2–4 business days, $9.95 (free over $199)
- Europe — 5–10 business days, $19.95 (free over $199)
- United States & Canada — 6–12 business days, $24.95 (free over $199)
- Rest of world — 10–20 business days, $49.95 (free over $499)
Free shipping is applied automatically at checkout once your order passes the threshold for your destination — no code needed. Every order ships with a tracking number emailed to you as soon as it leaves our warehouse.
International orders may be subject to import duties or customs fees set by your local authorities; these are not included in the checkout price and are the buyer’s responsibility.
Read our full Shipping Policy for details on processing, customs, damaged or lost parcels, and delivery addresses.
Related products
BOSGAME M5 AI Mini PC (Ryzen AI Max+ 395)
Price range: $3,299.99 through $3,599.99
Beelink EQi Wildcat Lake Core 3 304
Price range: $469.00 through $699.00
Beelink SEi14 AI Intel® Core™ Ultra 9 185H
Original price was: $2,199.00.$2,059.00Current price is: $2,059.00.
Beelink GTi15 Ultra Intel® Core™ Ultra 9 285H (OpenClaw Pre-Installed)
Price range: $1,499.00 through $1,999.00
Beelink SER10 MAX AMD Pro Ryzen™ AI 9 HX 470 (OpenClaw & Local LLM Pre-installed)
Price range: $1,339.00 through $2,299.00
Beelink SER8 AMD Ryzen™ 7 8845HS
Price range: $799.00 through $1,239.00
Beelink EQi13 Pro Intel® Core™ 13500H/13620H
Price range: $589.00 through $669.00
Anika J. –
Running Llama and Mixtral quantized locally as a coding assistant and this thing just gets on with it. The 32dB noise claim actually holds up under sustained load, which surprised me, and it’s effectively replaced my monthly API bill.
Noor A. –
Run this for local document processing tools at my small business so client data never touches a third-party cloud. It’s been reliable for months now with zero downtime.
Maya L. –
Set this up as both a homelab AI inference box and a network storage server using the dual 10GbE ports, and it handles both without breaking a sweat. For the unified memory you get versus buying an equivalent discrete GPU, the value here is hard to beat.
Hugo T. –
Had to flash a BIOS update before the NPU was properly recognised under Linux — not a huge deal once I found the right thread on the Beelink forum, but it would be nice if that worked out of the box. Runs great now.
Felix N. –
Using this for a mix of video editing and local AI work — the Radeon 8060S handles DaVinci Resolve timelines fine. Worth noting AMD’s ROCm software stack is still a bit behind CUDA for some of the newer AI frameworks, so check compatibility for your specific tools before buying if AI is the main use case.
Leila H. –
Arrived well packaged and was up and running on Windows 11 within the hour using LM Studio for local inference. Silent operation is real — I keep forgetting it’s on.
Adrian W. –
It’s a capable machine but the price is still a stretch for what is, on paper, a mini PC — even with these specs. Also had to manually grab the latest driver package from AMD’s site rather than everything working cleanly out of the box on Windows.
Owen C. –
Upgraded both M.2 slots and now run a 70B quantized model alongside my working dataset entirely on local storage. Build quality is solid. Fan noise is a touch more noticeable than the 32dB spec suggests once you’re actually pushing it hard, but still far from loud.
Zara Q. –
Went with this over the DGX Spark as a more budget-conscious way into local LLM development and honestly don’t regret it. Temperatures stayed reasonable even under sustained 140W load during longer fine-tuning runs.
Marek K. –
Been running this for about a week now on Ubuntu 24.04 and it’s been solid. With the 128GB unified memory you can allocate up to 96GB as VRAM, which is more than enough for the larger local models I’ve been testing. Idle power draw sits around 15W and climbs to roughly 180W under a full AI workload, so it’s not going to wreck your electricity bill even running overnight jobs. Only gripe is I’d like clearer guidance on BIOS updates before doing one myself. It’s bigger than a typical mini PC but for what’s inside, that’s expected. Delivery from ComputaHardware was quick, no complaints there.
Dario M. –
Picked this up to run as a dedicated local AI server and it’s been great once configured. I ended up running Lemonade as the inference backend and performance has been consistently good since. Setup took a bit of time — budget an evening for getting drivers and your stack sorted — but I actually enjoyed the process. Would recommend it to anyone comfortable doing a bit of tinkering.
Kevin T. –
Wasn’t expecting to see this at my door so soon — really happy with how quickly it shipped and arrived. Unit itself feels well built straight out of the box.
Sam R. –
Using this daily and it’s been reliable. Had one small intermittent issue in the first week and the support team sorted it out quickly — no back and forth, just a fast, complete fix. That kind of support makes a real difference when you’re relying on a machine for work.
Luca P. –
Really impressed with the build quality — it’s quiet even under load, which matters if it’s sitting on a desk. One thing worth flagging: the Linux NIC drivers aren’t included in the default kernel, so you’ll need to download them separately — not a big deal, just wasn’t expecting it. Support were responsive when I asked about it. Docking one star only for the extra driver step; otherwise an excellent machine.