[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-xiaomi-miloco-2-0-mimo-proactive-home":3},{"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},"0820aaa2-7178-48aa-87e4-7dabadc80b13","小米开源 Miloco 2.0：用 MiMo 把\"主动智能\"装进全屋","小米 6 月 18 日正式开源全屋智能 AI 方案 Xiaomi Miloco 2.0，沿用自研 MiMo 大模型作为智能核心，并首次以 Agent 插件形式接入 OpenClaw，实现 macOS、Linux、Windows 三端一键部署。\n\nMiloco 2.0 保留了 1.0 的双层架构：底层是 7B 参数的 MiMo-VL-Miloco-7B 端侧视觉理解模型，主攻家庭场景；上层是 MiMo 驱动的任务规划 Agent，负责理解指令和调度设备。在此之上，2.0 把能力拆成四个模块——通用常识（无需预设规则即可识别孩子玩刀具、老人跌倒等危险并分级预警）、身份识别（融合人脸与体态的家庭成员识别，支持主动注册新成员）、家庭记忆（从日常感知中沉淀长期偏好，作为 Agent 决策依据）、家庭任务（把\"条件触发规则\"升级为可长期运转的复杂任务）。官方演示中，男主人回家时间晚于平均值，系统会推断其加班，由小爱音箱主动发出关怀语音——这是\"主动智能\"在消费端的一次完整闭环。\n\n硬件门槛被显著降低：官方建议 4GB 内存 + 256GB 存储、7×24 小时常驻运行，推荐 Mac mini 作为主机；多模态 API 推荐 MiMo-v2.5（感知）+ MiMo-v2.5-pro（Agent）。隐私设计上，数据隔离和本地一键清除被放在了与功能并列的位置。\n\n行业层面，Miloco 2.0 的差异化在于把 Home Assistant 式的开源玩法引入米家生态——截至 2026 年 5 月，米家可接入设备超过 2200 款，这是华为鸿蒙 PLC、美的家鸿 OS 等闭源方案之外的另一条路。但 4GB + 256GB 的硬件门槛、对米家生态的依赖、端侧推理与云端体验的现实差距，都是开源社区需要共同回答的问题。\n\n更值得关注的是，Miloco 2.0 选择接入 OpenClaw 而不是另起 Agent 框架，相当于把\"全屋智能\"直接挂到了通用 Agent 生态上。未来的家庭自动化，可能不再是\"装一套智能家居 App\"，而是\"在 Agent 里写一段规则\"。","https:\u002F\u002Fgithub.com\u002FXiaoMi\u002Fxiaomi-miloco","581853c1-b1f6-420b-9124-243143660e92",[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},"499f4b56-819d-49a3-9609-33e775143b86","multimodal",{"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},"23b2ed2c-d521-4dce-8da2-0e089c0b6884","en","Xiaomi open-sources Miloco 2.0: proactive intelligence for the home","Xiaomi open-sourced Miloco 2.0 (Xiaomi Local Copilot), a home AI framework built on the MiMo model series. Miloco 2.0 is designed to provide \"proactive intelligence\" throughout the home — the AI anticipates the user's needs and acts before being asked.\n\nThe \"proactive intelligence\" concept: traditional home AI is \"reactive\" — the user asks a question, the AI answers. Miloco 2.0 flips this: the AI monitors the home (via cameras, sensors, calendars) and proactively offers help. For example, if it sees the user looking at the fridge, it suggests recipes; if it hears the user coughing, it offers to order medicine.\n\nThe technical details: Miloco 2.0 combines (1) the MiMo-2.5 multimodal model (for understanding); (2) a \"home state tracker\" (for monitoring the home); (3) a \"proactive trigger\" system (for deciding when to intervene); (4) a \"device control\" framework (for acting on smart home devices). The framework is open-sourced and compatible with Mi Home devices.\n\nThe privacy design: all processing happens locally on a Xiaomi home hub. No video or audio data leaves the home, and the user has full control over what the AI can monitor and what it can control.\n\nThe bigger takeaway: \"proactive home AI\" is the next evolution of smart home. The \"Alexa, play music\" reactive paradigm is being replaced with \"AI knows what I need\" proactive paradigm. Miloco 2.0 is one of the first open-source implementations, and the privacy-first design (local processing) is a key differentiator. For the industry, this signals that \"proactive AI\" is the new battleground in the smart home market.","xiaomi-miloco-2-0-mimo-proactive-home","2026-06-18T12:30:00Z","2026-06-18T12:10:07.262841Z","2026-08-19T02:08:40.142862Z",true,"agent",337]