[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-mimo-v2-5-openrouter":3,"news-related-37c8dc68-276b-4905-baed-1e120e86287c":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},"37c8dc68-276b-4905-baed-1e120e86287c","小米MiMo-V2.5登顶OpenRouter周榜月榜：国产开源MoE拿下全球调用量第一","7月27日,OpenRouter最新数据显示,小米MiMo-V2.5同时拿下全球大模型调用量周榜与月榜双第一,是当周唯一突破10T tokens的模型。从5月至今,其单周token量从1.46T攀升至10.46T,两个月增幅约616%。\n\n这一成绩之所以值得认真看,是因为OpenRouter是全球最大的LLM API聚合平台,榜单按真实调用量排序,不是benchmark刷分,意味着全球开发者用脚投票选出了它。MiMo-V2.5背后是MoE架构(总参过万亿、激活42B),混合注意力比7:1,1M上下文,定价仅Claude Opus 4.6的不到1\u002F3——性价比路线把开发者从闭源高价模型里抢了回来。\n\n放到更宽的坐标系看:2026上半年中国大模型已经不止在榜单上刷分,而是开始拿走真实流量。Hy3、Kimi K3、Qwen、MiMo轮番登顶OpenRouter。闭源前排模型在Coding Agent、Tool Use这些核心场景里依然领先(MiMo-V2-Pro的ClawEval 61.5 vs Opus 4.6的66.3),但调用量曲线已经反向。\n\n这不是国产替代的鸡汤故事。MiMo-V2.5跑赢的不是技术上限,而是技术性价比——用1\u002F8的推理成本做八成的能力。这条路如果被DeepSeek、Qwen复刻并扩张,闭源模型靠能力溢价撑高价的窗口,会比所有人预想的更早关闭。","https:\u002F\u002F36kr.com\u002Fnewsflashes\u002F3913798998201732","5e4fd3d1-9cb4-44a6-bae5-9ffb449c05c1",[10,14,17,20],{"id":11,"name":12,"slug":12,"description":13,"color":13},"120fa59a-ff6f-4537-9bf5-f818df636a0e","benchmark",null,{"id":15,"name":16,"slug":16,"description":13,"color":13},"a8002d98-9df1-4ab9-94d4-a7625af634c4","china-ai",{"id":18,"name":19,"slug":19,"description":13,"color":13},"01598627-1ea6-4b27-a5d8-874971571a71","llm",{"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":28},"4143b4e6-0536-423b-b2d6-464d16203781","en","Xiaomi MiMo-V2.5 tops OpenRouter: #1 global call volume","On July 27, the latest OpenRouter data shows that Xiaomi's MiMo-V2.5 simultaneously took the global LLM weekly and monthly call-volume double #1 — the only model that week to break 10T tokens. From May to date, its weekly token count climbed from 1.46T to 10.46T, about a 616% increase in two months. The reason this result is worth taking seriously: OpenRouter is the world's largest LLM API aggregation platform, with the leaderboard sorted by real call volume, not benchmark gaming — meaning global developers voted with their feet for it. MiMo-V2.5 is backed by a MoE architecture (over 1T total parameters, 42B activated), a 7:1 mixed attention ratio, 1M context, and pricing under one-third of Claude Opus 4.6 — the cost-performance route pulled developers back from the closed-source high-price model. Placed in a broader coordinate: in the first half of 2026, Chinese large models have stopped just gaming leaderboard scores and started capturing real traffic. Hy3, Kimi K3, Qwen, and MiMo have taken turns topping OpenRouter. Closed-source front-row models still lead in core scenarios like Coding Agent and Tool Use (MiMo-V2-Pro's ClawEval 61.5 vs Opus 4.6's 66.3), but the call-volume curve has reversed. This isn't a feel-good story of \"domestic substitution\". MiMo-V2.5 didn't beat the technical ceiling — it beat the technical price-performance ratio: 1\u002F8 of the inference cost, 80% of the capability. If this path gets replicated and scaled by DeepSeek and Qwen, the window for closed-source models to sustain high prices through capability premium will close earlier than anyone expects.","mimo-v2-5-openrouter","2026-07-28T03:00:00Z","2026-07-27T16:03:51.056716Z","2026-08-19T02:08:40.142862Z",true,"agent",311,{"items":37},[38,43,48,53,58,63],{"id":39,"title":40,"news_slug":41,"published_at":42},"ad3be632-49a1-44f5-9816-62c168e56467","全球大模型调用量榜前五全是\"中国造\":开源 MoE 正在重写 OpenRouter 的地理坐标","openrouter-top5-china-moe-open-source-2026w31","2026-08-02T03:30:00+00:00",{"id":44,"title":45,"news_slug":46,"published_at":47},"dfa4fd62-754c-4c08-b7c4-5f8bc4c7f678","上线一周调用量增 68 倍,腾讯混元 Hy3 在 OpenRouter 全球登顶","tencent-hunyuan-hy3-openrouter","2026-07-16T00:01:00+00:00",{"id":49,"title":50,"news_slug":51,"published_at":52},"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":54,"title":55,"news_slug":56,"published_at":57},"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":59,"title":60,"news_slug":61,"published_at":62},"68072ee1-fc37-4064-ab18-09550ae72d1b","GLM-5.3-Flash 把 320B MoE 跑在国产芯片上:Flash 价位和 $0.15 API 的混合注意力栈","glm-5-3-flash-chinese-chips-hybrid-attention","2026-08-27T03:00:00+00:00",{"id":64,"title":65,"news_slug":66,"published_at":67},"ff0bc92a-295a-4707-be8d-76115fe9eeee","PerceptionBench 出炉:16 个前沿多模态模型,视觉感知无一及格","moonshot-perceptionbench-atomic-perception","2026-08-26T13:15:00+00:00"]