[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-kimi-k3-launch-2-8t":3,"news-related-ebb562ad-9213-4db8-a29e-28dfba8df066":36},{"id":4,"title":5,"summary":6,"content":6,"original_url":7,"source_id":8,"tags":9,"translations":23,"news_slug":29,"published_at":30,"created_at":31,"modified_at":32,"is_published":33,"publish_type":34,"image_url":13,"view_count":35},"ebb562ad-9213-4db8-a29e-28dfba8df066","Kimi K3 上线:Moonshot 用 2.8 万亿参数与 KDA 线性注意力把开源带回牌桌","Moonshot AI 在 WAIC 开幕前夕上线 **Kimi K3**——2.8 万亿参数 MoE,比 DeepSeek V4 Pro(1.6T)大 75%,完整权重计划 7 月 27 日开源。亮点不只是堆参数:K3 首次把 Kimi Delta Attention(混合线性注意力 KDA)与 Attention Residuals 同时落地,叠加 FAST 2025 最佳论文 Mooncake 的 KV-cache 中心化推理栈。\n\n**性能侧 K3 直接撕掉\"开源追闭源\"叙事**:GDPval-AA v2 上 1687 分位列第三,落后 Claude Fable 5 Max(1815)与 GPT-5.6 Sol Max(1747.8);BrowseComp 长程检索 91.2 拿下 SOTA。**最关键演示是 48 小时自主芯片设计**:K3 无人干预跑通 4 mm²、100 MHz、8700 tokens\u002Fs 的自指芯片——把\"长程 Agent\"从榜单拉到可验证任务。\n\n**API $3 \u002F $15 每百万 token、缓存 $0.30**,兼容 OpenAI SDK,1M 上下文自动缓存;三档覆盖 256K–1M 窗口,迁移摩擦砍到接近零。\n\n开源追闭源三年,Kimi K3 用 **2.8 万亿 + 线性注意力 + 长上下文 + 完全开源 + OpenAI 兼容**一次堵上\"上不了生产\"的托词。护城河不再是参数,而是算法、推理栈与生态设计。","https:\u002F\u002Fplatform.kimi.ai\u002Fdocs\u002Fguide\u002Fkimi-k3-quickstart","0ec8f614-42c7-4256-8591-209e1e39eb6b",[10,14,17,20],{"id":11,"name":12,"slug":12,"description":13,"color":13},"a8002d98-9df1-4ab9-94d4-a7625af634c4","china-ai",null,{"id":15,"name":16,"slug":16,"description":13,"color":13},"01598627-1ea6-4b27-a5d8-874971571a71","llm",{"id":18,"name":19,"slug":19,"description":13,"color":13},"7e89b5cc-57db-4f37-bc6d-28919a73931c","model-release",{"id":21,"name":22,"slug":22,"description":13,"color":13},"b9bd9039-fcdb-41a8-b85b-fc1587def2b9","open-source",[24],{"id":25,"lang":26,"title":27,"summary":28,"content":13},"9a2d0605-d128-49cc-a675-d3711531775f","en","Kimi K3 arrives: 2.8T params bring open source back to the table","On the eve of WAIC, Moonshot AI put **Kimi K3** online — a 2.8-trillion-parameter MoE, 75% larger than DeepSeek V4 Pro (1.6T), with the full weights planned to be open-sourced on July 27. The highlights go beyond parameter count: K3 is the first to land Kimi Delta Attention (the hybrid linear-attention KDA) together with Attention Residuals, stacked on top of the FAST 2025 best-paper Mooncake's KV-cache-centric inference stack. On the performance side, K3 directly shreds the \"open-source chasing closed-source\" narrative: 1687 points on GDPval-AA v2, ranking third, behind Claude Fable 5 Max (1815) and GPT-5.6 Sol Max (1747.8); BrowseComp long-horizon retrieval 91.2, taking SOTA. The most critical demo is the 48-hour autonomous chip design: K3, without human intervention, completed a self-referential 4 mm², 100 MHz, 8700 tokens\u002Fs chip — pulling \"long-horizon Agent\" from leaderboard to a verifiable task. **API $3 \u002F $15 per million tokens, $0.30 for cache**, OpenAI-SDK-compatible, with 1M context auto-cached; three tiers cover the 256K–1M window, with migration friction cut close to zero. After three years of open-source chasing closed-source, Kimi K3 uses **2.8T + linear attention + long context + fully open source + OpenAI compatibility** to close all the \"can't get to production\" excuses at once. The moat is no longer parameter count, but algorithm, inference stack, and ecosystem design.","kimi-k3-launch-2-8t","2026-07-16T20:01:00Z","2026-07-16T20:06:46.980814Z","2026-08-19T02:08:40.142862Z",true,"agent",104,{"items":37},[38,43,48,53,58,63],{"id":39,"title":40,"news_slug":41,"published_at":42},"f6e4aab0-7693-4c2c-bb66-c1641fc2cc3e","Ox Alpha 谜底揭晓:智谱 GLM-5.3-Flash,MIT 开源 320B MoE","ox-alpha-glm-5-3-flash-reveal","2026-08-27T13:30:00+00:00",{"id":44,"title":45,"news_slug":46,"published_at":47},"804ab59a-a8d6-4b61-bf74-8f6f2bdae83c","智谱把 Flash 做成一件正经事:一次说清 GLM-5.3-Flash 的架构和 benchmark 真相","glm-5-3-flash-hybrid-attention-architecture","2026-08-27T08:00:00+00:00",{"id":49,"title":50,"news_slug":51,"published_at":52},"b0183d10-bcfd-44ed-a178-a2c813f10b69","国家超算互联网AI社区上线Kimi K3:2.8万亿参数MoE一键调用,开源大模型有了国产算力底座","kimi-k3-cnsc-internet-launch","2026-07-28T09:30:00+00:00",{"id":54,"title":55,"news_slug":56,"published_at":57},"3d8b9b1a-e038-466f-9b6b-304f911e35a7","Kimi K3 开源三件套 MoonEP\u002FFlashKDA\u002FAgentEnv:Moonshot 把 2.8T MoE 训练栈完整交底","kimi-k3-moonep-flashkda-agentenv","2026-07-28T04:30:00+00:00",{"id":59,"title":60,"news_slug":61,"published_at":62},"a151db0c-d832-4df2-ac03-2d4e58b26e99","Kimi K3 跑通 MiniTriton:Moonshot 让 LLM 第一次从零编译出自己的 GPU 编译器","kimi-k3-minitriton-gpu-compiler","2026-07-26T14:00:00+00:00",{"id":64,"title":65,"news_slug":66,"published_at":67},"cabef8bd-d6c3-429c-930a-6f1c51ddb0b4","华为开源 openPangu-2.0-Flash：92B\u002F6B MoE 把\"昇腾原生\"推到生产一线","huawei-openpangu-2-flash","2026-06-30T10:03:00+00:00"]