[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-a-lawyers-eight-used-v100s-run-deepseek-privately":10,"sections":45},{"siteName":4,"siteTagline":5,"publisherName":4,"contactEmail":6},"The Revision","Tech news, decoded.","editor@therevision.news",{"gaMeasurementId":8,"adsenseClientId":9},"G-ZW2MV82GYR","ca-pub-8533917693782264",{"article":11},{"id":12,"slug":13,"title":14,"dek":15,"body_md":16,"tags_json":17,"published_at":18,"created_at":19,"updated_at":20,"status":21,"review_note":22,"review_notes":23,"image_url":22,"persona_id":22,"persona_name":22,"section":34,"tags":35,"sources":40,"feedback":44,"feedback_at":22,"cost_usd":44,"total_tokens":44},12616,"a-lawyers-eight-used-v100s-run-deepseek-privately","A Lawyer's Eight Used V100s Run DeepSeek Privately","A homelab of eight 2018-era Nvidia GPUs runs a 248.8GB DeepSeek model at up to 215 tokens per second, but used V100 prices have more than doubled.","A lawyer has built a private AI server from eight used Nvidia Tesla V100 GPUs, so client case files never leave the office. According to the builder's Reddit post, it runs the open DeepSeek V4.1 Flash model at up to 215 tokens per second (a token is roughly a word fragment) on code and up to 118 on legal writing.\n\nThe cards are 2018-era data center parts, each with 32GB of fast memory. They sit on two four-card boards, and the cards on each board talk to one another over Nvidia's direct NVLink connection. The two boards only connect through the slower PCIe bus, so the builder split the model's layers in half, one half per board, to keep that hop to one per step. The model is compressed to mostly two bits per weight and still totals 248.8GB. About 110GB of lookup tables sit in ordinary RAM, which the builder says barely affects speed.\n\nThe speeds come with limits. At 128K tokens of input, the first word of the answer takes 59.2 seconds, and the server handles one request at a time. Three days before the post, four cards managed only 36 to 38 tokens per second. The builder credits the second board plus custom software work for the jump.\n\nStability took effort. The host reset twice with a PCIe error, which stopped after the builder capped GPU clocks. Each card now runs at 150W, half its rating. The post also mentions a burn mark on one board's power connector, and memory errors that cleared when the RAM was slowed to DDR4-2666.\n\nThe builder estimates 50 to 60 percent of existing workflows can now run locally instead of on frontier models. Nothing goes out without human review.\n\nThe cheap-hardware route is closing, though. The four-card boards cost about $2,400 each, but used cards now list for about $900, against roughly $400 in the post, and prices appear to be rising. The R580 driver branch is also the last that supports the V100's Volta architecture, so the hardware is both getting pricier and approaching the end of its software support.","[\"local-ai\",\"nvidia\",\"homelab\",\"deepseek\"]","2026-11-09T13:20:00.000Z","2026-11-09T14:52:30.755Z","2026-11-09T14:52:30.933Z","published",null,[24,30],{"id":25,"reviewer":26,"round":27,"reason":28,"status":29},"editor-r1","editor",1,"The headline's 'Offline' and the lead's claim that client files 'never leave the building' overstate the source: the builder says the setup will protect client data 'once it is self contained, which is the goal,' and it currently hands work off to a second DeepSeek server on DGX Sparks. Drop 'offline' from the headline and dek, state the privacy goal as a goal rather than an achieved fact, and note that the $2,400-per-board price covers more than the four cards (4 × $400 = $1,600).","resolved",{"id":31,"reviewer":26,"round":32,"reason":33,"status":29},"editor-r2",2,"The 'offline' overstatement is fixed, but two problems remain: the closing line calls the V100s 'cheap memory' when it means the cheap GPUs or hardware, and the draft says the builder 'removed a board with a burned power connector' while the lead says all eight cards on two boards are running, so clarify what was removed (the source is ambiguous) or drop the claim.","ai",[36,37,38,39],"local-ai","nvidia","homelab","deepseek",[41],{"name":42,"url":43},"Tom's Hardware","https:\u002F\u002Fwww.tomshardware.com\u002Ftech-industry\u002Fartificial-intelligence\u002Flawyer-builds-private-ai-server-with-eight-used-nvidia-v100-gpus-up-to-215-tokens-per-second-to-keep-legal-files-offline-costing-about-usd2-400-per-four-gpu-board",0,{"sections":46},[47,51,56,61,66,71,76,81,86,91,96,101,106,111],{"name":48,"slug":34,"count":49,"latest_published_at":50},"AI",7144,"2026-11-12T13:27:45.000Z",{"name":52,"slug":53,"count":54,"latest_published_at":55},"Security","security",1128,"2026-11-11T18:05:00.000Z",{"name":57,"slug":58,"count":59,"latest_published_at":60},"Deals","deals",642,"2026-11-12T14:00:00.000Z",{"name":62,"slug":63,"count":64,"latest_published_at":65},"Policy","policy",619,"2026-11-12T12:53:36.000Z",{"name":67,"slug":68,"count":69,"latest_published_at":70},"Consumer Tech","consumer-tech",364,"2026-11-12T11:12:32.000Z",{"name":72,"slug":73,"count":74,"latest_published_at":75},"Hardware","hardware",336,"2026-11-12T11:24:43.000Z",{"name":77,"slug":78,"count":79,"latest_published_at":80},"Science","science",230,"2026-11-12T10:00:00.000Z",{"name":82,"slug":83,"count":84,"latest_published_at":85},"Software","software",168,"2026-11-11T17:36:49.000Z",{"name":87,"slug":88,"count":89,"latest_published_at":90},"Startups","startups",159,"2026-11-11T12:30:00.000Z",{"name":92,"slug":93,"count":94,"latest_published_at":95},"Dev Tools","dev-tools",127,"2026-11-11T19:00:44.000Z",{"name":97,"slug":98,"count":99,"latest_published_at":100},"General","general",118,"2026-11-11T16:49:42.000Z",{"name":102,"slug":103,"count":104,"latest_published_at":105},"Gaming","gaming",107,"2026-11-12T10:09:03.000Z",{"name":107,"slug":108,"count":109,"latest_published_at":110},"Reviews","reviews",38,"2026-10-20T12:00:00.000Z",{"name":112,"slug":113,"count":114,"latest_published_at":115},"How-To","how-to",8,"2026-10-05T09:00:00.000Z"]