Google has announced the Google Home Speaker, a new smart speaker built around its Gemini AI assistant. It goes on sale June 25 at 16,800 yen and comes in two colors, Hazel and Porcelain. In place of the Google Assistant used by earlier Google Nest models, the speaker is designed from the ground up for Gemini for Home, which lets you control devices throughout the house through natural conversation. The biggest change is that your instructions get through in the way you would normally speak, without memorizing fixed phrases.

A smart speaker that drops the memorized commands

Until now, smart speakers generally required you to learn set commands such as "OK Google, turn on the lights." The new Google Home Speaker changes that premise entirely. The voice assistant inside has been replaced with Gemini for Home, which uses advanced language understanding and reasoning to grasp your intent based on how you actually speak.

In other words, there is no need to memorize an incantation to operate the device. You can simply say what comes to mind, and even a roundabout phrasing is interpreted by Gemini, which then acts on it. For anyone who has felt that smart homes are hard to approach, being able to handle appliances as an extension of ordinary conversation lowers the barrier to entry.

Handling multiple requests, slips of the tongue, and context

The new model's intelligence shows in how it handles complicated requests. For example, it can process an instruction that spans conditions, such as "turn off all the lights except the bedside lamp." You can also bundle several tasks into a single breath, like "dim the kitchen lights, play some relaxing music, and set a 20-minute timer."

It is flexible about misstatements, too. If you correct yourself by saying "turn off the coffee maker... oh, I meant turn it on," it cancels the earlier "off" command and switches it on correctly. It can also handle a deeper level of reasoning: ask "what will the weather be like for the next baseball game?" and it works out the date and location of the match before returning a pinpoint forecast for that day and place.

Because it remembers the context of a conversation for a short while, you can have a genuine back-and-forth. There is no need to restate the premise each time when you want to dig into a question or string follow-up requests together. On top of that, the Continued Conversation feature keeps the microphone on for a brief moment after a response, so you can skip "OK Google" from the second sentence onward. This is the first time the feature has been available in every supported language, including Japanese.

Hardware tuned for conversation, plus 360-degree sound

To make natural dialogue work, the hardware has been carefully built as well. The processor is a quad-core A55 (2.0GHz) with a built-in NPU, so AI processing is handled quickly on the device itself. For picking up sound, it carries three high-sensitivity microphones that capture your voice accurately even in a noisy room. On the privacy side, it has a physical two-stage microphone mute switch, allowing you to cut the mic at the hardware level.

Its abilities as a speaker have also been raised. Google says the balanced 360-degree sound delivers audio clearly no matter where in the room it sits. It supports Voice Match, which distinguishes who is speaking. In addition, by linking it with the Google TV Streamer television device, you can pair up to two speakers to build a home theater with spatial surround support.

For connectivity, it supports Wi-Fi 6 (IEEE 802.11ax), Bluetooth 5.4, and a Thread 1.3 border router. The unit measures 107mm in diameter and 86.6mm in height, and weighs 396g. The cable is a direct 1.5m type. As an environmental consideration, more than 37 percent of the whole product is made from recycled materials.

A subscription extends it into an AI that watches over the home

Natural voice control works even for free, but signing up for the paid Google Home Premium (from 1,000 yen per month) broadens what Gemini can do. With Gemini Live, for instance, you only need to say "OK Google, let's talk" to enjoy freer dialogue, such as brainstorming, changing the subject midway, or interrupting.

In homes that combine it with a Google Nest Cam, the speaker can tell you about movement the camera has caught and what is happening right now. Ask "is the back gate open?" and you can check whether you forgot to close it; ask "did the pet get on the sofa today?" and you can look back over the day's movements. You can also use Daily Summary, which has Gemini sum up what happened around the house while you were away. The speaker thus expands its role from a tool you operate by voice into a companion that keeps watch over the home and reports back.

Summary

The new Google Home Speaker takes a step from a smart speaker that makes you memorize fixed commands toward an AI speaker you can address in everyday words. It handles multiple instructions, misstatements, and reasoning that spans context, and Continued Conversation now works in Japanese as well. At an accessible 16,800 yen, it adds 360-degree sound and home theater pairing, and adding a subscription extends it all the way to home monitoring and free-flowing dialogue. It looks set to be an easy-to-grasp entry point for figuring out how to use Gemini inside the home.