以 GEEKOM 列出的每部 3,999 美元計算,四部 A9 Mega 約需 15,996 美元,未計額外儲存、網絡設備、部署、電費及支援成本。
目前公告未交代網絡拓撲、量化格式、分片方式或實際 tokens per second,因此這次示範較像部署概念驗證,而非已獲證實的生產級方案。
How did GEEKOM demonstrate running the roughly 284-billion-parameter DeepSeek V4 Flash Mixture-of-Experts model locally and without cloud inAn editorial visualization of GEEKOM’s four-node A9 Mega local AI cluster.
AI 提示
Create a landscape editorial hero image for this Studio Global article: How did GEEKOM demonstrate running the roughly 284-billion-parameter DeepSeek V4 Flash Mixture-of-Experts model locally and without cloud in. Article summary: GEEKOM’s demonstration was a four-node, local distributed-inference setup: it linked four A9 Mega mini PCs over USB4, then used Ubuntu, AMD ROCm, and DwarfStar software to partition an optimized DeepSeek V4 Flash model a. Topic tags: general, general web, user generated, academic. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, char
openai.com
GEEKOM 最近展示了一個四節點的本地 AI 推理平台:將四部 A9 Mega 迷你電腦透過 USB4 連成一個分散式叢集,再配合 Ubuntu、AMD ROCm 及 DwarfStar,將 DeepSeek V4 Flash 分拆到不同節點上運行,最後透過 OpenAI-compatible API 提供服務。
換句話說,企業內部應用程式、AI agent 或開發工具,可以像連接一般 OpenAI 風格端點一樣,呼叫這個留在本地網絡內的模型。GEEKOM 將方案定位為毋須依賴數據中心或雲端推理的 private AI 部署。
以 GEEKOM 列出的每部 3,999 美元計算,四部 A9 Mega 約需 15,996 美元,未計額外儲存、網絡設備、部署、電費及支援成本。
目前公告未交代網絡拓撲、量化格式、分片方式或實際 tokens per second,因此這次示範較像部署概念驗證,而非已獲證實的生產級方案。
How did GEEKOM demonstrate running the roughly 284-billion-parameter DeepSeek V4 Flash Mixture-of-Experts model locally and without cloud inAn editorial visualization of GEEKOM’s four-node A9 Mega local AI cluster.
AI 提示
Create a landscape editorial hero image for this Studio Global article: How did GEEKOM demonstrate running the roughly 284-billion-parameter DeepSeek V4 Flash Mixture-of-Experts model locally and without cloud in. Article summary: GEEKOM’s demonstration was a four-node, local distributed-inference setup: it linked four A9 Mega mini PCs over USB4, then used Ubuntu, AMD ROCm, and DwarfStar software to partition an optimized DeepSeek V4 Flash model a. Topic tags: general, general web, user generated, academic. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, char
openai.com
GEEKOM 最近展示了一個四節點的本地 AI 推理平台:將四部 A9 Mega 迷你電腦透過 USB4 連成一個分散式叢集,再配合 Ubuntu、AMD ROCm 及 DwarfStar,將 DeepSeek V4 Flash 分拆到不同節點上運行,最後透過 OpenAI-compatible API 提供服務。
換句話說,企業內部應用程式、AI agent 或開發工具,可以像連接一般 OpenAI 風格端點一樣,呼叫這個留在本地網絡內的模型。GEEKOM 將方案定位為毋須依賴數據中心或雲端推理的 private AI 部署。