{"id":389905,"date":"2026-08-19T09:32:32","date_gmt":"2026-08-19T08:32:32","guid":{"rendered":"https:\/\/droix.net\/blogs\/?p=365004"},"modified":"2026-09-27T03:38:47","modified_gmt":"2026-09-27T02:38:47","slug":"geekom-predstavlja-cetveroslozni-ai-klaster-mini-racunala-s-lokalnim-deepseek-modelom","status":"publish","type":"post","link":"https:\/\/droix.net\/blogs\/hr\/geekom-predstavlja-cetveroslozni-ai-klaster-mini-racunala-s-lokalnim-deepseek-modelom\/","title":{"rendered":"GEEKOM predstavlja \u010detveroslo\u017eni AI klaster mini ra\u010dunala s lokalnim DeepSeek modelom"},"content":{"rendered":"\n

GEEKOM<\/a> predstavio je \u010detveroslo\u017eni AI klaster izgra\u0111en od svojih kompaktnih A9 Mega Mini ra\u010dunala, pokazuju\u0107i kako se sustavi malog formata mogu kombinirati za lokalne poslovne AI primjene.<\/p>\n\n\n\n

Klaster povezuje \u010detiri A9 Mega<\/a> jedinice putem USB4, omogu\u0107uju\u0107i im rad kao distribuirana platforma bez potrebe za tradicionalnim serverskim ormari\u0107em ili vlasni\u010dkom mre\u017enom opremom velike brzine. Svaki sustav pokre\u0107e AMD-ov Ryzen AI Max+ 395 <\/a>procesor, s 16 Zen 5 CPU jezgri, Radeon 8060S grafikom i objedinjenom memorijom.<\/p>\n\n\n\n

\n
\"\u010cetiri<\/a>
\u010cetiri A9 Mega jedinice<\/figcaption><\/figure>\n\n\n\n
\"\u010cetiri<\/a>
\u010cetiri A9 Mega jedinice<\/figcaption><\/figure>\n<\/figure>\n\n\n\n

Pokre\u0107u\u0107i Ubuntu, ROCm i DwarfStar softver, klaster distribuira DeepSeek V4 Flash model na sva \u010detiri stroja. Tako\u0111er pru\u017ea API kompatibilan s OpenAI-jem, omogu\u0107uju\u0107i integraciju s aplikacijama, AI agentima i alatima za automatizaciju.<\/p>\n\n\n\n

Klju\u010dna prednost ovog postava je lokalna obrada. Tvrtke mogu analizirati dokumente, izvorni kod, ugovore i interno znanje bez slanja osjetljivih informacija u javni oblak. GEEKOM ka\u017ee da je sustav tako\u0111er testiran s kontekstnim prozorima do 250.000 tokena, \u0161to ga \u010dini prikladnim za duga\u010dke dokumente i slo\u017eene radne tokove.<\/p>\n\n\n\n

\n
\"\u010cetiri<\/a>
\u010cetiri GEEKOM A9 Mega jedinice<\/figcaption><\/figure>\n\n\n\n
\"\u010cetiri<\/a>
\u010cetiri A9 Mega jedinice, pogled straga<\/figcaption><\/figure>\n<\/figure>\n\n\n\n

Prema tvrtki, klaster je postigao pribli\u017eno 14,61 tokena u sekundi pri pojedina\u010dnoj istovremenosti, s 95. percentilom vremena do prvog tokena od oko 0,42 sekunde u testovima s 32 i 128 tokena. Organizacije mogu zapo\u010deti s jednim ili dva A9 Mega<\/a> sustava i pro\u0161iriti se na \u010detiri kako njihovi zahtjevi rastu.<\/p>\n\n\n\n

Demonstracija isti\u010de potencijal kompaktnih ra\u010dunala kao skalabilnih lokalnih AI platformi za tvrtke, laboratorije, u\u010dionice i okru\u017eenja rubnog ra\u010dunalstva.<\/p>\n\n\n\n

Izvor: GEEKOM putem PR Newswire<\/a>. Podaci o performansama i mogu\u0107nosti temelje se na testiranju koje je prijavila tvrtka.<\/em><\/p>\n","protected":false},"excerpt":{"rendered":"

GEEKOM predstavio je \u010detveroslo\u017eni AI klaster izgra\u0111en od svojih kompaktnih A9 Mega Mini ra\u010dunala. Povezan putem USB4 i pokretan AMD Ryzen AI Max+ 395 procesorima, sustav lokalno pokre\u0107e DeepSeek V4 Flash, nude\u0107i tvrtkama skalabilan na\u010din obrade osjetljivih dokumenata, koda i AI radnih optere\u0107enja bez potpunog oslanjanja na cloud infrastrukturu.<\/p>\n","protected":false},"author":859,"featured_media":389906,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[6500,8518],"tags":[30227,8751,6540],"class_list":["post-389905","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-novosti","category-mini-pc-vijesti","tag-a9-mega-12","tag-geekom","tag-mini-racunalo"],"lang":"hr","translations":{"hr":389905,"en":365004,"de":370301,"en-gb":370672,"en-ie":370679,"en-au":370686,"fr":371387,"ar":382327,"bg":384272,"cs":385446,"es":387053,"da":387374,"pt-pt":388137,"hu":390629,"lv":393581,"el":403888,"et":403954,"fi":404020,"it":404085,"lt":404150,"nl":404259,"no":404325,"pl":404390,"ro":404456,"sk":404522,"sl":404588,"sv":404653},"pll_sync_post":{},"_links":{"self":[{"href":"https:\/\/droix.net\/blogs\/wp-json\/wp\/v2\/posts\/389905","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/droix.net\/blogs\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/droix.net\/blogs\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/droix.net\/blogs\/wp-json\/wp\/v2\/users\/859"}],"replies":[{"embeddable":true,"href":"https:\/\/droix.net\/blogs\/wp-json\/wp\/v2\/comments?post=389905"}],"version-history":[{"count":1,"href":"https:\/\/droix.net\/blogs\/wp-json\/wp\/v2\/posts\/389905\/revisions"}],"predecessor-version":[{"id":389911,"href":"https:\/\/droix.net\/blogs\/wp-json\/wp\/v2\/posts\/389905\/revisions\/389911"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/droix.net\/blogs\/wp-json\/wp\/v2\/media\/389906"}],"wp:attachment":[{"href":"https:\/\/droix.net\/blogs\/wp-json\/wp\/v2\/media?parent=389905"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/droix.net\/blogs\/wp-json\/wp\/v2\/categories?post=389905"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/droix.net\/blogs\/wp-json\/wp\/v2\/tags?post=389905"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}