[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$f2uja89k9qizax":3},{"slug":4,"category":5,"publishedAt":6,"titleZh":7,"titleEn":8,"summaryZh":9,"summaryEn":10,"models":11,"hardwares":12,"bodyZh":13,"bodyEn":14,"sourceName":15,"sourceUrl":16,"updatedAt":17},"xiaomi-mimo-v2-6-open","NEWS","2026-10-01T05:05:00.000Z","小米开源 MiMo-V2.6：Pro 版登顶 Artificial Analysis 开放权重榜","Xiaomi Open-Sources MiMo-V2.6: Pro Tops the Artificial Analysis Open-Weight Board","小米发布并开源 MiMo-V2.6 系列（完全开放权重）：分 Pro \u002F Flash \u002F Ultraspeed 三档并附桌面客户端，全系原生全模态；据行业日报报道，Pro 版登顶 Artificial Analysis 开放权重模型榜，分数高于 Kimi K3 与 Qwen3.8 Max。","Xiaomi has released and open-sourced the MiMo-V2.6 series with fully open weights: three tiers (Pro \u002F Flash \u002F Ultraspeed) plus a desktop client, all natively omni-modal. Industry digests report the Pro tier tops the Artificial Analysis open-weight leaderboard, ahead of Kimi K3 and Qwen3.8 Max.",[],[],"## 发布\n\n据行业日报报道，小米于 2026 年 9 月 22 日正式发布并开源 **MiMo-V2.6 系列**大模型，**完全开放权重**——普通开发者可以自行部署，进一步压低成本，「不用再盯着国外的 Llama、Mistral」。\n\n## 三个档位\n\n- **Pro**：面向复杂项目\n- **Flash**：面向普通场景\n- **Ultraspeed**：面向极致速度场景\n- 连桌面客户端都直接提供\n\n全系列**原生全模态**；报道推测其在端侧推理效率上做了特别优化（媒体推测，非官方口径）。\n\n## 榜单与成本\n\n- 据报道，**Pro 版登顶 Artificial Analysis 开放权重模型榜**，分数高于 Kimi K3 与 Qwen3.8 Max\n- API 沿用此前定价，单任务成本约 **$0.13**——约为此前同类旗舰的三分之一\n- 小米将其描述为「探索递归自我改进路径的关键成果」\n\n## 尚未披露\n\n- 参数规模：报道未提及\n- 具体开源许可证名称：报道未提及\n- 上下文长度：报道未提及\n\n> 开放权重阵营在 9 月竞争白热化：DeepSeek V4.1 Flash（9 月 10 日）刚坐上开放权重榜首，MiMo-V2.6 Pro（9 月 22 日）即完成接棒。对本站读者而言，这意味着本地可部署模型的能力天花板仍在快速抬升。","## Launch\n\nAccording to industry daily reports, Xiaomi officially released and open-sourced the **MiMo-V2.6 series** on September 22, 2026, with **fully open weights** — independent developers can self-deploy to cut costs further, no longer \"having to watch Llama and Mistral from abroad.\"\n\n## Three Tiers\n\n- **Pro**: for complex projects\n- **Flash**: for general scenarios\n- **Ultraspeed**: for maximum-speed scenarios\n- A desktop client ships alongside\n\nThe whole series is **natively omni-modal**; reports speculate special optimizations for edge-side inference efficiency (media speculation, not an official claim).\n\n## Leaderboard and Cost\n\n- The **Pro tier reportedly tops the Artificial Analysis open-weight leaderboard**, scoring above Kimi K3 and Qwen3.8 Max\n- API pricing unchanged, at roughly **$0.13 per task** — about one-third of comparable flagships\n- Xiaomi describes it as \"a key milestone on the path of recursive self-improvement\"\n\n## Not Yet Disclosed\n\n- Parameter count: not mentioned in reports\n- Specific open-source license name: not mentioned\n- Context length: not mentioned\n\n> September saw white-hot competition in open weights: DeepSeek V4.1 Flash (Sep 10) had just taken the open-weight crown when MiMo-V2.6 Pro (Sep 22) took it over. For our readers, the capability ceiling of locally deployable models keeps rising fast.","AI 日报（5x10.cn）","https:\u002F\u002F5x10.cn\u002Fpost\u002F619.html","2026-10-01T05:14:10.763Z"]