[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"article-ai-weekly-2026-w25-zh":3,"article-related-ai-weekly-2026-w25-zh":27,"series-industry-6cb42a56-614e-43f9-8258-ccd76cdcfa9f":70},{"id":4,"slug":5,"title":6,"content":7,"summary":8,"source":9,"source_url":10,"author":11,"image_url":12,"cover_image":12,"category":13,"language":14,"translated_content":11,"related_article_id":15,"keywords":16,"key_takeaways":11,"views":24,"created_at":25,"published_at":26,"topic_cluster_id":11},"6cb42a56-614e-43f9-8258-ccd76cdcfa9f","ai-weekly-2026-w25-zh","AI 週報：2026-06-08 ~ 2026-06-15","\u003Cp>本週主旋律很清楚：模型跑得更快、入口更貼近日常，但推理可靠度還沒跟上。從小米 MiMo 的高吞吐模型到 Gemini 進入 Maps，產業在把 AI 從展示品推向可用介面；同時，機率題研究提醒我們，能力表現仍高度依賴題型與提示方式。\u003C\u002Fp>\u003Cp>另一條線則是工程化落地。微調、代碼審查、容器工作流與 AI factory 都在說同一件事：真正的競爭不只在模型大小，而在部署、成本、流程與算力供應是否能接住需求。\u003C\u002Fp>\u003Ch2>趨勢雷達\u003C\u002Fh2>\u003Ctable>\u003Ctr>\u003Cth>維度\u003C\u002Fth>\u003Cth>信號\u003C\u002Fth>\u003Cth>本週變化\u003C\u002Fth>\u003Cth>影響\u003C\u002Fth>\u003C\u002Ftr>\u003Ctr>\u003Ctd>模型\u003C\u002Ftd>\u003Ctd>強\u003C\u002Ftd>\u003Ctd>小米 MiMo-V2.5-Pro-UltraSpeed 把 1T 模型推到最高 1000 tokens\u002Fs。\u003C\u002Ftd>\u003Ctd>高吞吐會把「能不能用」改寫成「用多少成本、多久回本」。\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd>Agent\u003C\u002Ftd>\u003Ctd>中\u003C\u002Ftd>\u003Ctd>Google Gemini 進一步接入 Maps，Ask Maps 與導航對話化更完整。\u003C\u002Ftd>\u003Ctd>Agent 會更像內建工作流，而不是獨立聊天視窗。\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd>開源\u003C\u002Ftd>\u003Ctd>弱\u003C\u002Ftd>\u003Ctd>Linux 7.1-rc7 顯示 AMD Zen 6 支援持續成熟。\u003C\u002Ftd>\u003Ctd>開源生態先行完成硬體適配，會影響後續伺服器與工作站採購節奏。\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd>算力與基建\u003C\u002Ftd>\u003Ctd>中\u003C\u002Ftd>\u003Ctd>韓國與 Nvidia 對話聚焦 26 萬顆晶片、AI factory 與首爾研發中心。\u003C\u002Ftd>\u003Ctd>主權 AI 與區域算力中心會把供應鏈競爭推向國家級配置。\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd>應用\u003C\u002Ftd>\u003Ctd>強\u003C\u002Ftd>\u003Ctd>Gemini 在 Google Maps 變成對話介面，RAG 與微調的分工也被重新討論。\u003C\u002Ftd>\u003Ctd>應用層會更重視場景適配，文風、流程與地圖導航這類任務將分流處理。\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd>政策與監管\u003C\u002Ftd>\u003Ctd>平靜\u003C\u002Ftd>\u003Ctd>無重大進展\u003C\u002Ftd>\u003Ctd>監管本週沒有新錨點，但估值與披露壓力仍會回到 IPO 與大模型商業化討論。\u003C\u002Ftd>\u003C\u002Ftr>\u003C\u002Ftable>\u003Ch2>重點事件\u003C\u002Fh2>\u003Ch3>小米把 1T 模型推到 1000 tokens\u002Fs\u003C\u002Fh3>\u003Cp>\u003Cstrong>發生什麼。\u003C\u002Fstrong> 小米在 \u003Ca href=\"\u002Fnews\u002Fxiaomi-mimo-1t-model-1000-tokens-per-second-zh\">MiMo-V2.5-Pro-UltraSpeed\u003C\u002Fa> 上把 1T 級模型推到最高 1000 tokens\u002Fs，並同步調整定價與舊模型退場時程。\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781497229891-ck4n.png\" alt=\"AI 週報：2026-06-08 ~ 2026-06-15\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781497225186-pksn.png\" alt=\"AI 週報：2026-06-08 ~ 2026-06-15\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>\u003Cstrong>為什麼重要。\u003C\u002Fstrong> 這不是單純跑分，而是把大模型從「可展示」推進到「可計費、可排程、可部署」的工程狀態。當速度與價格一起變動，市場會開始用吞吐和單位成本評估模型，而不是只看參數量。\u003C\u002Fp>\u003Cp>\u003Cstrong>誰受影響、下一步觀察。\u003C\u002Fstrong> 受影響的是雲端推理平台、終端廠商與企業採購方；接下來要看的是實際延遲、長上下文穩定性，以及新定價是否真的壓縮同級模型的使用門檻。\u003C\u002Fp>\u003Ch3>Gemini 進入 Google Maps，導航開始對話化\u003C\u002Fh3>\u003Cp>\u003Cstrong>發生什麼。\u003C\u002Fstrong> Google 把 Gemini 更深地塞進 \u003Ca href=\"\u002Fnews\u002Fgoogle-gemini-latest-update-maps-zh\">Google Maps\u003C\u002Fa>，主打 Ask Maps 與 Immersive Navigation，讓地圖查詢與路線規劃更像對話。\u003C\u002Fp>\u003Cp>\u003Cstrong>為什麼重要。\u003C\u002Fstrong> 地圖是高頻入口，一旦自然語言成為主要操作方式，AI 就不再只是回答問題，而是直接改寫使用者操作介面。這類整合比單獨推出聊天產品更容易累積黏性，也更容易把搜尋、導航與本地服務串成一條鏈。\u003C\u002Fp>\u003Cp>\u003Cstrong>誰受影響、下一步觀察。\u003C\u002Fstrong> 受影響的是通勤族、旅遊用戶、在地商家與地圖 API 生態；下一步看 Google 是否把更多查詢、訂位與推薦動作接到同一個對話流程裡。\u003C\u002Fp>\u003Ch3>研究提醒：LLM 在反直覺機率題會失手\u003C\u002Fh3>\u003Cp>\u003Cstrong>發生什麼。\u003C\u002Fstrong> 研究指出，LLM 在標準機率題表現不錯，但遇到反直覺、改寫或帶誤導提示的題目時，準確率會明顯下滑。\u003C\u002Fp>\u003Cp>\u003Cstrong>為什麼重要。\u003C\u002Fstrong> 這說明模型的「看起來會」和「穩定會」之間仍有差距，尤其在推理題、決策輔助與教育場景裡，題目措辭會直接影響輸出品質。對產品方來說，不能只看 benchmark 平均值，還要測邊界案例。\u003C\u002Fp>\u003Cp>\u003Cstrong>誰受影響、下一步觀察。\u003C\u002Fstrong> 受影響的是做推理型產品的團隊、教育應用與評測平台；下一步要看是否有更針對反直覺題型的評測集與訓練方法出現。\u003C\u002Fp>\u003Ch3>微調與 RAG 的分工被重新釐清\u003C\u002Fh3>\u003Cp>\u003Cstrong>發生什麼。\u003C\u002Fstrong> 文章指出，當目標是文風而不是事實時，\u003Ca href=\"\u002Fnews\u002Ffine-tuning-beats-rag-style-not-facts-zh\">微調比 RAG 更有效\u003C\u002Fa>；RAG 比較適合補事實，不適合改寫作風格。\u003C\u002Fp>\u003Cp>\u003Cstrong>為什麼重要。\u003C\u002Fstrong> 這件事看似技術細節，實際上影響產品架構與成本。很多團隊把所有問題都交給 RAG，結果在語氣、格式與品牌口吻上效果有限；把任務拆清楚，才能少走冤枉路。\u003C\u002Fp>\u003Cp>\u003Cstrong>誰受影響、下一步觀察。\u003C\u002Fstrong> 受影響的是內容生成、客服、自動寫作與企業知識助理；下一步看哪些團隊會把「事實補強」與「風格控制」拆成兩條管線。\u003C\u002Fp>\u003Ch2>下週觀察\u003C\u002Fh2>\u003Cul>\u003Cli>關注 \u003Cstrong>Google\u003C\u002Fstrong> 是否在 \u003Cstrong>Maps\u003C\u002Fstrong> 之後，把 Gemini 更深接到搜尋與本地商家流程。\u003C\u002Fli>\u003Cli>觀察 \u003Cstrong>小米 MiMo-V2.5-Pro-UltraSpeed\u003C\u002Fstrong> 的實測延遲、定價落地與舊模型退場是否按表推進。\u003C\u002Fli>\u003Cli>留意 \u003Cstrong>Nvidia 與韓國\u003C\u002Fstrong> 的 AI factory 談判，特別是 26 萬顆晶片交付與首爾研發中心進度。\u003C\u002Fli>\u003Cli>檢查 \u003Cstrong>Linux 7.1\u003C\u002Fstrong> 對 AMD \u003Cstrong>Zen 6\u003C\u002Fstrong> 的支援是否進一步進入主線整合。\u003C\u002Fli>\u003Cli>追蹤是否有新的 \u003Cstrong>LLM 推理評測\u003C\u002Fstrong>，專門測反直覺機率題與提示改寫魯棒性。\u003C\u002Fli>\u003C\u002Ful>","本週主旋律是模型效率與應用入口同步前進：小米把 1T 模型推到 1000 tokens\u002Fs，Google 讓 Gemini 進入 Maps，研究則提醒 LLM 在反直覺推理上仍會失手。","oracore.dev","https:\u002F\u002Foracore.dev\u002Fnews\u002Fai-weekly-2026-w25-zh",null,"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781497229891-ck4n.png","industry","zh","56930ca4-3142-46e2-8c5d-a837b6de8651",[17,18,19,20,21,22,23],"AI 週報","趨勢雷達","人工智慧","科技新聞","大模型效率","AI 應用","AI 基建",0,"2026-06-15T04:00:28.945351+00:00","2026-06-15T04:00:28.935+00:00",{"tags":28,"relatedLang":29,"relatedPosts":33},[],{"id":15,"slug":30,"title":31,"language":32},"ai-weekly-2026-w25-en","AI Weekly: 2026-06-08 ~ 2026-06-15","en",[34,40,46,52,58,64],{"id":35,"slug":36,"title":37,"cover_image":38,"image_url":38,"created_at":39,"category":13},"0d168fc7-0d4b-4653-aba4-1f058a075b7d","midjourney-v8-1-default-model-update-zh","Midjourney V8.1 變成預設模型，速度與細節都升級","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781515078543-4z93.png","2026-06-15T09:17:18.754939+00:00",{"id":41,"slug":42,"title":43,"cover_image":44,"image_url":44,"created_at":45,"category":13},"856138b5-19e2-4328-9637-ca9baa17e48f","midjourney-vs-zh","Midjourney 免費方案 vs 付費方案","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781514187435-4dch.png","2026-06-15T09:02:34.997559+00:00",{"id":47,"slug":48,"title":49,"cover_image":50,"image_url":50,"created_at":51,"category":13},"4784f345-852e-4293-96fe-aa51d1b45522","anthropic-35b-buildout-finance-chips-zh","Anthropic 的 350 億美元擴建證明，AI 已經是金融與晶片的戰場","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781510578626-4jiq.png","2026-06-15T08:02:22.25572+00:00",{"id":53,"slug":54,"title":55,"cover_image":56,"image_url":56,"created_at":57,"category":13},"6b67d939-3742-448d-9198-fe8263c61bfd","openai-partner-network-enterprise-ai-access-zh","OpenAI Partner Network 擴大企業導入","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781506997762-yo0j.png","2026-06-15T07:02:32.038955+00:00",{"id":59,"slug":60,"title":61,"cover_image":62,"image_url":62,"created_at":63,"category":13},"6f47f82d-e595-48f5-a60c-cc68ec965c38","anthropics-offline-move-turns-policy-into-code-zh","Anthropic 下線把政策變成程式","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781488986698-jrh4.png","2026-06-15T02:02:39.436181+00:00",{"id":65,"slug":66,"title":67,"cover_image":68,"image_url":68,"created_at":69,"category":13},"ac93db8b-1365-40f0-a324-df636352ed06","mcp-server-directory-for-tool-builders-zh","89.2k 星 MCP 伺服器目錄，先找再接最省時","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781482670481-y6y5.png","2026-06-15T00:17:20.984744+00:00",[71,76,81,86,91,96,101,106,111,116],{"id":72,"slug":73,"title":74,"created_at":75},"ee073da7-28b3-4752-a319-5a501459fb87","ai-in-2026-what-actually-matters-now-zh","2026 AI 真正重要的事","2026-03-26T07:09:12.008134+00:00",{"id":77,"slug":78,"title":79,"created_at":80},"83bd1795-8548-44c9-9a7e-de50a0923f71","trump-ai-framework-power-speech-state-preemption-zh","川普 AI 框架瞄準電力、言論與州權","2026-03-26T07:12:18.695466+00:00",{"id":82,"slug":83,"title":84,"created_at":85},"ea6be18b-c903-4e54-97b7-5f7447a612e0","nvidia-gtc-2026-big-ai-announcements-zh","NVIDIA GTC 2026 重點拆解","2026-03-26T07:14:26.62638+00:00",{"id":87,"slug":88,"title":89,"created_at":90},"4bcec76f-4c36-4daa-909f-54cd702f7c93","claude-users-spreading-out-and-getting-better-zh","Claude 用戶更分散，也更會用","2026-03-26T07:22:52.325888+00:00",{"id":92,"slug":93,"title":94,"created_at":95},"bd903b15-2473-4178-9789-b7557816e535","openclaw-raises-hard-question-for-ai-models-zh","OpenClaw 逼問 AI 模型價值","2026-03-26T07:24:54.707486+00:00",{"id":97,"slug":98,"title":99,"created_at":100},"eeac6b9e-ad9d-4831-8eec-8bba3f9bca6a","gap-google-gemini-checkout-fashion-search-zh","Gap 把結帳搬進 Gemini","2026-03-26T07:28:23.937768+00:00",{"id":102,"slug":103,"title":104,"created_at":105},"0740e53f-605d-4d57-8601-c10beb126f3c","google-pushes-gemini-transition-to-march-2026-zh","Google 把 Gemini 轉換延到 2026 年 3…","2026-03-26T07:30:12.825269+00:00",{"id":107,"slug":108,"title":109,"created_at":110},"e660d801-2421-4529-8fa9-86b82b066990","metas-llama-4-benchmark-scandal-gets-worse-zh","Meta Llama 4 分數風波又擴大","2026-03-26T07:34:21.156421+00:00",{"id":112,"slug":113,"title":114,"created_at":115},"183f9e7c-e143-40bb-a6d5-67ba84a3a8bc","accenture-mistral-ai-sovereign-enterprise-deal-zh","Accenture 攜手 Mistral AI 賣主權 AI","2026-03-26T07:38:14.818906+00:00",{"id":117,"slug":118,"title":119,"created_at":120},"191d9b1b-768a-478c-978c-dd7431a38149","mistral-ai-faces-its-hardest-year-yet-zh","Mistral AI 迎來最硬的一年","2026-03-26T07:40:23.716374+00:00"]