Google Home gets upgraded Gemini voice assistant and new camera controls

Google Home 迎来 Gemini 语音助手升级与全新摄像头控制功能

Google launched its big AI-fueled redesign of Google Home late last year, and it has been adding features here and there ever since. Today, the company announced a bigger update that might take care of some of your smart home woes. Camera feeds will be easier to navigate, and the AI event labeling should be more straightforward. The move to Gemini 3.1 for Home voice assistance should also mean the robot is less obtuse and more reliable.

Google 在去年年底推出了以 AI 为核心的 Google Home 重大改版，此后一直在不断增加新功能。今天，该公司宣布了一项更重大的更新，旨在解决用户在使用智能家居时遇到的一些痛点。摄像头画面将更易于浏览，AI 事件标记也将更加直观。语音助手升级至 Gemini 3.1 后，其响应将不再那么生硬，可靠性也会有所提升。

According to Google, Home users who have signed up for the early access channel should already have the update to Gemini 3.1. Google initially released this AI model on other platforms in February, but that rollout didn’t include Google’s smart speakers. With the expansion to Home, Google says those speakers will be able to take advantage of Gemini 3.1’s “advanced reasoning to better interpret and execute complex, multi-step voice commands.” Of course, it says something like that with every Gemini update.

据 Google 称，已注册抢先体验频道的 Home 用户应该已经收到了 Gemini 3.1 的更新。Google 最初于 2 月在其他平台上发布了该 AI 模型，但当时的推送并不包含 Google 的智能音箱。随着此次扩展至 Home，Google 表示这些音箱将能够利用 Gemini 3.1 的“高级推理能力，更好地理解并执行复杂的多步骤语音指令”。当然，每次 Gemini 更新时他们都会这么说。

Google has cited various AI evaluations that show Gemini 3.1 is better at parsing big, complex prompts. It showed gains in tests like ARC-AGI-2 and Humanity’s Last Exam, both of which require tricky logic problems that need domain-specific knowledge. How much that kind of capability will benefit a smart speaker that specializes in brief interactions is unclear, but you can have long conversations with Gemini in your smart home devices if you want. Google notes the improved model can process multiple different tasks in a single prompt, saving you from breaking up tasks into multiple commands.

Google 引用了多项 AI 评估结果，显示 Gemini 3.1 在解析大型、复杂提示词方面表现更佳。它在 ARC-AGI-2 和“人类最后一场考试”（Humanity’s Last Exam）等测试中均有提升，这些测试都需要处理涉及特定领域知识的复杂逻辑问题。这种能力对主要用于简短交互的智能音箱有多大帮助尚不明确，但如果你愿意，现在可以在智能家居设备上与 Gemini 进行长对话。Google 指出，改进后的模型可以在单个提示词中处理多个不同任务，无需你将任务拆分为多条指令。

The AI-powered Ask Home feature is also expanding beyond the app with this update. In the near future, the Home-specific Gemini chatbot will be available in the Home web interface, allowing you to conversationally check camera history and create automations. This will start as a preview feature, though.

随着此次更新，AI 驱动的“Ask Home”功能也将走出 App。在不久的将来，专为 Home 设计的 Gemini 聊天机器人将登陆 Home 网页端，让你能够通过对话方式查看摄像头历史记录并创建自动化流程。不过，该功能初期将作为预览版推出。

Google is also adding new automation options, both triggers and actions. Here’s the full list:

Security & Access Control: Arm and disarm systems, door lock monitoring (locked, unlocked, jammed, forced open, or ajar), and binary sensors (contact, leak, freeze).
Appliance & Cleaning: Appliance state control (Start, Stop, Pause, Resume) and specific commands for robot vacuums.
Lighting & Environment Control: Brightness, On/Off, light effects, color/temperature changes, window covering positions, and thermostat humidity monitoring.
Media & Entertainment: Playback control and volume management.
Device Health & Inputs: Battery/charging monitoring and smart switch inputs (presses, long presses, release events).

Google 还增加了新的自动化选项，涵盖触发器和动作。完整列表如下：

安全与访问控制： 系统的布防与撤防、门锁状态监控（锁定、解锁、卡住、被强行打开或未关严）、二进制传感器（接触/无接触、漏水/无漏水、结冰/无结冰）。
家电与清洁： 家电状态控制（启动、停止、暂停、恢复）以及扫地机器人的特定指令。
照明与环境控制： 亮度调节、开关、灯光效果、颜色/色温更改、窗帘位置控制以及恒温器的湿度监控。
媒体与娱乐： 播放控制和音量管理。
设备健康与输入： 电池/充电监控以及智能开关输入（按压、长按、释放事件）。

Paying subscribers can use Ask Home to create automations (with or without the new ingredients), but everyone can do it the old-fashioned way with buttons and drop-down menus.

付费订阅用户可以使用 Ask Home 创建自动化流程（无论是否使用上述新功能），但所有用户依然可以通过传统的按钮和下拉菜单方式进行设置。

Less camera jank

减少摄像头操作的卡顿感

The 2025 Home app update made a lot of changes, but it didn’t exactly nail the experience. Camera control was still annoying, and the AI notifications weren’t very helpful (or accurate). Today’s update could address at least some of that, but Google says the underlying model is not being upgraded to Gemini 3.1. That’s only for the voice assistant end. Google says timeline navigation will be much improved now, featuring a higher frame rate when scrubbing through video. That should make it easier to get where you’re trying to go, and if you miss it, there are finally simple skip buttons that take you 10 seconds forward or backward. There are also new swipe controls for switching between timeline/event views and resizing the player window.

2025 年的 Home App 更新做了很多改动，但体验并不完美。摄像头控制依然令人烦恼，AI 通知也不够实用（或准确）。今天的更新至少解决了一部分问题，但 Google 表示底层模型并未升级到 Gemini 3.1，该升级仅针对语音助手端。Google 表示，时间轴导航现在得到了显著改善，在拖动视频进度时帧率更高。这让你更容易定位到想要查看的时间点；如果错过了，现在终于有了简单的快进/快退 10 秒按钮。此外，还增加了新的滑动控制功能，用于在时间轴/事件视图之间切换以及调整播放器窗口大小。

With the move to Gemini, Google Home’s ability to recognize people fell off considerably, making notifications much less useful. Home will now have thumbs up/down feedback buttons, which will help make Familiar Face notifications more accurate over time. Facial recognition will also automatically ignore blurry or obscured images to improve IDs. Notifications and timeline previews will also have better animated preview GIFs that are centered on the subject, too.

随着转向 Gemini，Google Home 的人脸识别能力大幅下降，导致通知的实用性大打折扣。现在，Home 增加了“点赞/点踩”反馈按钮，这将有助于随着时间的推移提高“熟悉面孔”通知的准确性。人脸识别功能现在还会自动忽略模糊或被遮挡的图像，以提高识别率。通知和时间轴预览也将提供更好的动画 GIF 预览，且画面会自动以主体为中心。

The AI model that analyzes camera footage in Google Home has a tendency to mislabel things, which can lead to frustrating and misleading notifications. Google isn’t claiming a major change to the quality of descriptions, but the system has apparently been streamlined to make the descriptions simpler. Google says this means you’ll see less cluttered and more straightforward event labels that are “more accurate.” We will see about that.

Google Home 中用于分析摄像头画面的 AI 模型容易出现误标，这会导致令人沮丧且具有误导性的通知。Google 并未声称描述质量有重大改变，但系统显然经过了精简，使描述更加简洁。Google 表示，这意味着你将看到更简洁、更直观且“更准确”的事件标签。我们拭目以待。

You won’t get Ask Home or the AI event descriptions with a free Home profile. But those paying for extended footage will see the allegedly simpler AI-generated descriptions in more places. Descriptions will come to the timeline view, as well as to older Nest cameras that previously didn’t work with Gemini. You have to make sure Gemini for Home features are enabled in the camera settings, though.

免费的 Home 账户无法使用 Ask Home 或 AI 事件描述功能。但付费购买扩展视频存储服务的用户将在更多地方看到这些据称更简洁的 AI 生成描述。这些描述将出现在时间轴视图中，并支持此前无法使用 Gemini 的旧款 Nest 摄像头。不过，你必须确保在摄像头设置中启用了“Gemini for Home”功能。