$16,000 Quad AI Geekom mini PC cluster gets DeepSeek V4 Flash treatment with 512GB RAM — reaches 14.61 tokens per second

Four GEEKOM A9 Mega mini PCs link together through simple USB4 cables Each unit runs on the AMD Ryzen AI Max+ 395 processor chip Combined cluster memory reaches 512GB across all four connected systems Geekom has released the DeepSeek V4 Flash model across four A9 Mega mini PCs, forming a distributed

<![CDATA[ <article> <ul><li><strong>Four GEEKOM A9 Mega mini PCs link together through simple USB4 cables</strong></li><li><strong>Each unit runs on the AMD Ryzen AI Max+ 395 processor chip</strong></li><li><strong>Combined cluster memory reaches 512GB across all four connected systems</strong></li></ul><p>Geekom has <a href="https://www.prnewswire.com/apac/news-releases/geekom-pushes-the-boundaries-of-mini-pc-innovation-with-four-node-a9-mega-ai-cluster-302852825.html" target="_blank" rel="nofollow">released</a> the DeepSeek V4 Flash model across four A9 Mega mini PCs, forming a distributed cluster aimed at enterprise AI workloads.</p><p>The setup connects the devices through USB4 rather than relying on a traditional <a href="https://www.techradar.com/pro/best-data-center-proxies">data center</a> server.</p><p>Each A9 Mega runs on the AMD Ryzen AI Max+ 395 chip, which combines 16 Zen 5 <a href="https://www.techradar.com/news/best-processors">CPU</a><strong> </strong>cores with Radeon 8060S graphics and unified memory in one compact chassis</p><h2 id="local-processing-over-cloud-dependence">Local processing over cloud dependence</h2><p>The four-node configuration brings a combined 512GB of RAM to the task, with Ubuntu, ROCm, and DwarfStar software distributing the optimized model across the machines.</p><p>An OpenAI-compatible API links applications and AI agents to the cluster, while USB4 removes any need for a proprietary switch or server rack.</p><p>The arrangement allows organizations to keep prompts, documents, source code, and credentials within local infrastructure rather than routing them through a public cloud.</p><p>Businesses could theoretically build private knowledge assistants capable of searching contracts, manuals, and internal reports without external exposure.</p><p>For agent-based systems, Geekom says the cluster can process tools, policies, memory, and logs before executing any action, with a technical path that has operated using contexts up to 250K tokens.</p><h2 id="performance-numbers-under-scrutiny">Performance numbers under scrutiny</h2><p>Testing reported by Geekom recorded approximately 14.61 tokens per second at single concurrency, while P95 time to first token reached about 0.42 seconds.</p><p>The company says the configuration also provides greater capacity for long prompts, rather than concentrating solely on faster short-response generation.</p><p>Users can begin with one or two A9 Mega systems before expanding the configuration to four nodes as workloads increase over time.</p><p>Each machine can operate independently, while connected systems can contribute to distributed inference when greater computing capacity becomes necessary for demanding workloads.</p><p>The A9 Mega <a href="https://www.techradar.com/best/mini-pcs">mini PC</a> itself supports up to 128GB of LPDDR5x memory, although four systems provide the stated 512GB cluster capacity for distributed inference workloads.</p><p>Its specifications include 120W sustained performance, dual M.2 PCIe Gen4x4 storage slots, and support for up to 8TB of RAID-configured storage.</p><p>Geekom <a href="https://www.geekompc.com/geekom-a9-mega-ai-mini-pc/">lists the A9 Mega at $3,999</a> for a configuration with 128GB RAM and 2TB SSD, while the four-system cluster reaches $16,000.</p><p>The individual units also carry 126 total TOPS of combined processing output, with 120W of sustained thermal design power under load.</p><p>It also supports Wi-Fi 7, Bluetooth 5.4, dual 2.5G Ethernet, and four simultaneous 8K displays for professional environments.</p><p>The performance values above are vendor-supplied benchmarks without third-party testing - therefore, the real-world reliability of the setup for sustained enterprise use has not been independently confirmed. </p><figure class="van-image-figure inline-layout" data-bordeaux-image-check ><div class='image-full-width-wrapper'><div class='image-widthsetter' style="max-width:676px;"><p class="vanilla-image-block" style="padding-top:31.51%;"><img id="diM9tpwF2Lz85R8q85CT78" name="tr-g_news" alt="Google logo on a black background next to text reading 'Click to follow TechRadar'" src="https://cdn.mos.cms.futurecdn.net/diM9tpwF2Lz85R8q85CT78.jpg" mos="" align="middle" fullscreen="" width="676" height="213" attribution="" endorsement="" class="inline"></p></div></div></figure> </article> ]]>