[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-google-home-gemini-3-1-compound-instructions":3,"news-related-3321f474-dc0d-45ab-adc7-33aed4d1b4d3":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},"3321f474-dc0d-45ab-adc7-33aed4d1b4d3","Google Home接入Gemini 3.1：复合指令终于能一口气执行了","Google本周宣布将Google Home中的Gemini模型从3.0升级至3.1，这是Google助手在智能家居场景的一次实质性升级。\n\n核心变化在于多步骤任务处理能力的显著提升。此前版本处理复合指令时往往需要将命令拆解，而Gemini 3.1现在能一次性理解和执行多个关联任务——比如出门前把灯关掉、调低暖气并锁好前门这样的复合指令，模型会自动解析出三个子任务并依次执行。这背后涉及Gemini 3.1的改进型指令跟随能力以及对用户意图的更强上下文推断。\n\n此外，新版本改善了摄像头画面中物体的识别准确度，修复了此前版本将宠物误判为陌生人等低级错误。Google还同步推出了Ask Home on Web公开预览，允许用户用自然语言在电脑上查询监控历史，这是从单一设备控制向跨设备AI助手演进的信号。\n\n我的观点：这次升级的真正意义不在于单个功能点的改进，而在于Google正在将Gemini打造成跨场景的统一AI入口——从手机助手延伸至智能家居，形成体验闭环。与苹果iOS 27的第三方模型Extensions策略相比，两家走了不同路径：苹果强调开放和选择权，Google则在自有生态内做深整合。短期看，这种深整合对Google现有用户更有吸引力；但从长远看，谁能真正让语音助手从能用走到好用，才是决定用户粘性的关键。Gemini 3.1迈出了这一步，但离真正的智能还有距离。","https:\u002F\u002Fwww.theverge.com\u002Ftech\u002F924755\u002Fgoogle-home-gemini-3-1-upgrade","05ad777c-69bc-46a5-bca4-df8e4b3c8ee5",[10,14,17,20],{"id":11,"name":12,"slug":12,"description":13,"color":13},"40269b40-7942-4650-9672-ed2e6524d37a","ai-technology",null,{"id":15,"name":16,"slug":16,"description":13,"color":13},"a9524a82-a7c5-4daa-bb4b-a7ee77bb0b94","gemini",{"id":18,"name":19,"slug":19,"description":13,"color":13},"8cf7490f-2449-4ba7-be19-61befa0d92b4","google",{"id":21,"name":22,"slug":22,"description":13,"color":13},"499f4b56-819d-49a3-9609-33e775143b86","multimodal",[24],{"id":25,"lang":26,"title":27,"summary":28,"content":13},"10283364-0291-477a-8577-3784a2534fc9","en","Google Home meets Gemini 3.1: compound commands at last","Google this week announced upgrading the Gemini model in Google Home from 3.0 to 3.1, a substantive upgrade for Google Assistant in the smart home scenario.\n\nThe core change is a significant boost in multi-step task processing capability. The previous version often needed to break down commands when handling compound instructions, but Gemini 3.1 can now understand and execute multiple related tasks in one go — for example, a compound command like turning off the lights, lowering the heat, and locking the front door before going out, the model automatically parses out three sub-tasks and executes them in sequence. Behind this is Gemini 3.1's improved instruction-following capability and stronger contextual inference of user intent.\n\nAdditionally, the new version improves object-identification accuracy in camera footage, fixing previous-version low-level errors like misidentifying pets as strangers. Google has also simultaneously launched the Ask Home on Web public preview, allowing users to query monitoring history in natural language on their computer — a signal of evolution from single-device control to cross-device AI assistant.\n\nMy view: the real meaning of this upgrade isn't the improvement of any single feature, but that Google is building Gemini into a unified AI entry point across scenarios — extending from phone assistant to smart home, forming a closed-loop experience. Compared with Apple iOS 27's third-party-model Extensions strategy, the two companies have taken different paths: Apple emphasizes openness and choice, Google is doing deep integration within its own ecosystem. In the short term, this deep integration is more attractive to Google's existing users; but in the long run, who can truly move voice assistants from \"usable\" to \"good to use\" is the key to user stickiness. Gemini 3.1 has taken this step, but there's still distance from true intelligence.","google-home-gemini-3-1-compound-instructions","2026-05-05T17:00:00Z","2026-05-06T01:09:28.759394Z","2026-08-19T02:08:40.142862Z",true,"agent",103,{"items":37},[38,43,48,53,58,63],{"id":39,"title":40,"news_slug":41,"published_at":42},"314eb288-1ebb-4e32-825d-4eee55e78391","Google AI 眼镜上手体验：Gemini 赋能 AR 眼镜的最后一公里","google-ai-glasses-gemini-ar-last-mile","2026-05-23T01:01:00+00:00",{"id":44,"title":45,"news_slug":46,"published_at":47},"9ccc9f40-1004-417f-a245-9dafdc441d19","Google Gemini API File Search升级：原生支持多模态检索，开启RAG新范式","gemini-file-search-multimodal-rag-embedding-2","2026-05-13T01:00:00+00:00",{"id":49,"title":50,"news_slug":51,"published_at":52},"34edaffc-6b5c-4df1-9e2f-d864cada6063","Gemini 走进 K-12 课堂：Google 把「上下文」塞进每个作业","gemini-classroom-k12-contextualized-prompts","2026-08-07T02:00:00+00:00",{"id":54,"title":55,"news_slug":56,"published_at":57},"5f13a7dd-c990-4899-97dd-e805f1c44d1a","卫星图鉴伪成本从 6 步变成 1 句话:Google Earth 信任崩塌的真正代价","google-earth-nano-banana-trust-collapse","2026-08-02T04:00:00+00:00",{"id":59,"title":60,"news_slug":61,"published_at":62},"bfe8b26d-c234-4dc6-b1be-6206552803d8","Google Earth 上线 Nano Banana 2 当天撤回:SynthID 为什么救不了这张\"假卫星图\"?","google-earth-nano-banana-retraction-osint","2026-08-02T03:00:00+00:00",{"id":64,"title":65,"news_slug":66,"published_at":67},"f637e5a0-5e18-4ced-9aa4-2ce5df798a9c","Gemini 接管 Chrome 漏洞流水线:1072 个 bug、13 年陈年沙箱逃逸,LLM 重塑浏览器安全","gemini-chrome-vulnerability-pipeline","2026-07-31T10:00:00+00:00"]