I’m wearing a pair of thick-rimmed glasses on my face. They don’t feel heavy, but they feel chunky. I walk over to a poster of a painting—Girl With a Pearl Earring—and ask out loud what was so special about it. A brief answer detailing its expert use of light and color by Johannes Vermeer floats into my ears, and when I ask a follow-up about when it was painted, I quickly hear the same voice say, “around 1665.”

I’m not talking to myself, I swear. Nor am I hearing imaginary voices. No, I’m wearing a prototype of Google’s upcoming smart glasses, powered by its Gemini voice assistant. The company teased these smart glasses at its I/O developer conference earlier this year, showing a proof-of-concept video of AI-powered smart glasses using the name Project Astra. The pair I gazed through and chatted with uses that same Astra technology, but here it’s been built into a functioning product. Even though the glasses are still in their development phase, Google plans to release them sometime in 2025.

These smart glasses are one part of Google’s big announcement today: Android XR. This “extended reality” platform marks the 10th year of Google’s mobile operating system expanding to new platforms beyond phones, joining the ranks of Wear OS, Google TV, and Android Auto. It sets the stage for a new wave of virtual and augmented reality headsets and glasses with a customized version of Android running on them.

Most products running Android follow the same blueprint; Google manages the operating system, but various companies sell devices running the Android software. However, Android XR is a close collaboration between Google and Samsung (plus Qualcomm), much like the Wear OS smartwatch platform. It will debut with a Samsung mixed-reality headset codenamed Project Moohan, which also has a 2025 launch window. I got a chance to demo this headset alongside the glasses.

Android XR was announced today at a developer event in New York City that’s meant to get app builders interested in the platform ahead of the official product releases. My demos were with early versions of hardware and software, which is why I was not allowed to capture any photos or videos of my experience.

Android XR

Android XR arrives nearly a decade after Samsung and Google launched their respective smartphone-powered virtual reality platforms—Gear VR and Daydream View. Both companies shuttered their VR ambitions a few years later, but Shahram Izadi, vice president of AR and XR at Google, assures me things are different this time around.

“The technology is at an inflection point,” Izadi says. Generative AI “has brought about a need for this kind of form factor with new use cases that were hard to predict back then and because of some of the breakthroughs that we’ve had. Then there’s the developer on-ramping and the sort of platform approach rather than just a singular product. These things give us confidence that this is different.” He also points out that Google and Samsung have both continued the development of AR and VR technologies even after their initial platforms shut down.

The platform approach is important because any Android app will be accessible in an Android XR face computer. If an app has a tablet-optimized experience for larger screens, then it’ll be even more functional. Naturally, for a true spatially designed experience, developers will need to opt in and do some legwork, but Google says it’s possible to build for Android XR using many of the same tools currently used to develop Android apps. Android XR also supports OpenXR, Unity, and WebXR to ensure wider compatibility with apps and experiences.

But what makes Android XR different from anything else and prior versions is that it’s developed with Google’s Gemini assistant at its core. When you activate Gemini in a mixed reality headset or smart glasses, it will be able to hear and see what you’re seeing. For example, you can ask Gemini to rearrange the apps in the virtual environment if everything’s cluttered instead of dragging these apps yourself. I also witnessed Gemini narrating a YouTube video (of a soccer game) in real time as I watched. Let me jump into my experiences to explain further.

Project Moohan

Samsung’s mixed-reality headset looks nice, though very much like a standard VR headset. It feels a bit plasticky, but it’s relatively lightweight. I tightened it via a familiar dial on the back of the headrest, and while I only spent close to 30 minutes with the device on, it didn’t feel too cumbersome on my head. (My brow was sweaty though.) Taking a page from Apple’s Vision Pro, Samsung has a stand-alone battery pack powering the headset, meaning you’ll have to put this battery in your pocket, and there will be a wire running between the two pieces.

People needing corrective lenses will be able to get prescription inserts that stick to the headset’s pancake lenses. There’s also an optional light blocker that quickly attaches to the bottom of the headset in case you want a more immersive experience. Samsung had some ready-made inserts for my prescription available so I was able to remove my glasses, and I also opted to use it with the light blocker—it did a pretty good job of shutting out the ambient light, though there was some slight bleed around my nose.

While the interface will support VR controllers, the default way to interact with everything is your hands. In this way, it’s very similar to the Vision Pro. Use your hands to move the cursor around and pinch your index finger and thumb to select. You can pinch and zoom in or out to scale items, scroll, and so on. The headset also supports eye tracking, though this requires an additional bit of calibration: Once set up, just gaze at an icon and make the pinch gesture to select it. There’s no need to raise your arm. You can use both simultaneously and it feels responsive. With six degrees of freedom, you can move about a room in a virtual environment, and you’re protected by virtual fences in case you walk too close to a wall.

The visual quality of the display is sharp, though things get a little pixelated in the color-see-through mode. Like most headsets, Project Moohan uses foveated rendering; the part of the image at the center of your vision is the sharpest and rendered with the best quality, with the image quality dropping off toward your periphery. I never felt nauseous or dizzy (this has rarely been a problem for me in VR experiences), but I did feel as though things in the center of the screen sometimes felt too close together. This could just be an issue related to my interpupillary distance, even though the headset did detect my IPD and used the eye-tracking cameras to calibrate it. (I was told eye-tracking data never leaves the device.)

I opened up various Google apps, resized them, and placed them around the room. Text was sharp in apps like Chrome, and 3D models pulled from Google Search looked highly detailed. Google says it’s optimizing its apps for spatial environments, so YouTube, as an example, deconstructed its interface to have the video front and center, the video description on the left with comments, and related videos on the right.

Google Photos is also getting a neat trick: Any photo can be viewed like normal in 2D, or you can view them spatially, in three dimensions. It doesn’t matter whether the image was captured with a film camera, a smartphone, or a professional mirrorless camera. Google says it’s using machine learning models to understand the apparent depth in images and videos, allowing you to get that more immersive 3D media experience without having to capture spatial content (though spatial capture seems like a feature that will land on Pixel phones in the future, just like the capability was added to iPhones once Vision Pro arrived).

I watched a 2D video of a little boy interacting with a newborn, but then I tapped the immersive button and the video suddenly gained more depth and felt more lifelike. Some edges around the subjects were jagged, kind of like when portrait mode on your phone camera doesn’t do a great job deciding where the blurring should begin. But even though it was clear there was some processing happening, it still looked impressive. That said, I’m still not going to put on a VR headset to relive my memories (yet).

Naturally, you can work in a mixed-reality environment with a connected Bluetooth keyboard and mouse, and you can put yourself in an immersive environment if you want to focus, or leave see-through mode turned on to make sure your coworkers aren’t taking photos and giggling while you wear a ridiculous headset to get stuff done. It wasn’t clear if you’d be able to connect the headset to a laptop to bring your work into mixed reality, a feature available on the Apple Vision Pro.

Gemini in XR

A tap on the side of the headset brings up an app launcher, and this is where you can toggle on Gemini if you want it to persistently stay “on.” Once it’s on, there’s an icon at the top of the virtual space so that you are aware that everything you say and look at is being registered by Gemini.

In see-through mode, you can walk up to an object and ask Gemini about it—a Googler demoing the headset (before I tried it) walked up to someone else wearing an FC Barcelona shirt and asked Gemini to find the “standings of this team.” Gemini quickly registered the team name and pulled up search results with league standings and scores from recent matches.

You can ask Gemini anything like this and it will answer with visual results displayed in the headset. I asked it to “take me to Peru,” and it opened up a 3D version of Google Maps. I was able to move around and center on Lima, and in cities where Maps already has a lot of 3D models, you can explore areas in greater detail. You can keep talking to Gemini in these experiences, so I asked questions such as when would be the best time to visit and got a prompt answer.

In another example, I peeked inside a restaurant in New York City to take a virtual tour of the space. Google says it can use AI to stitch together images of a venue’s interior and display it so that it feels like you’re there. It did a pretty good job, and I asked Gemini if the place takes reservations, without having to specifically say the name, because I was staring at the name of the restaurant. It does take reservations, but Gemini couldn’t actually make one for me. (That integration might come later.)

Next, I watched a few videos on YouTube, where 2D content looks sharp and colorful. Stereoscopic content was even better; my senses felt surrounded. I watched some hikers walking along a trail and asked Gemini where this all was, and it said, “New Zealand.” I wasn’t able to verify that, but it looked like the right answer. I watched some more spatialized playback of 2D videos as the virtual player added depth and layering to make them feel 3D. I hopped over to the Google TV app and enabled a “Cinema mode” to launch a virtual theater for watching movies and shows, just like on other VR headsets.

Circle to Search, the feature Google debuted earlier this year on Android phones, is also available in Android XR. Just walk up to a physical object near you, press the top button on the headset, and then pinch and draw a circle around the thing you want to know more about. You’ll get a Google Search page with results.

Smart Glasses

Project Moohan very much feels like Google and Samsung catching up to the rest of the VR market, though the Gemini integration gives their efforts a unique layer. However, I will admit I was far more excited to try the smart glasses, where Gemini feels like it could be even more helpful. They didn’t disappoint. I walked over to another room and there were several pairs of glasses in front of me. Some were sunglasses, others had clear lenses. Like the headset, you can get them loaded up with your prescription. Google did not provide a name for the prototype glasses.

The glasses have thick arms, but they’re not too different from something like the Meta Wayfarers. Several years ago, Google bought North, a company that made smart glasses, and the design influence is apparent. These new frames are decidedly slimmer than North’s Focals and comfier to wear—you just have to get used to the thick rim around your eyes. There’s a camera built into the frame, with an LED that turns on when the camera is active.

The bulk of the processing is handled on your smartphone, which you pair to the frames during setup. Offloading much of the computing to your phone is what allows the glasses’ battery to last “all day,” or so Google hopes. Unlike Meta’s glasses, Google’s smart glasses have a display. And there might be options to choose from—glasses without a display at all but with access to Gemini, glasses with a monocular display on one lens, or glasses with a binocular display that shows up front and center for the best experience, each presumably increasing in price.

A tap on the side of the frame activates the display, bringing up general information like the date, time, and weather. Then you can launch apps. I got an early look at Google Maps, where, when looking ahead, you’re able to see turn-by-turn directions overlaid in the real world. Look down and you’ll get a visual of the Google Maps app kind of like you’re looking down at a phone. It gives you a bird’s-eye view of your current navigation in case you want to see the full route. (There will be other app integrations in the future, Google says.)

The star of the show is Gemini (powered by Gemini 2.0), which you can enable and pause with a tap on the glasses’ arm. This is what I used when I asked Gemini to explain the importance of the painting I was staring at and it gave me a succinct response. Next, I was staring at a road sign in Spanish. I asked Gemini to translate it, and the translated text showed up overlaid on top of the sign. Even crazier was when Google had someone stand in front of me and speak a sentence in Spanish. Their words were translated into English in real time on the glasses, allowing me to understand them without finagling with an app on my phone.

At last year’s Google I/O, Google teased smart glasses that could caption real-world conversations, a feature that’s helpful for people with hearing loss. I got a chance to see this in action during my demo. The experience wasn’t perfect, but it was good enough to follow a conversation, and frankly just a little mind-boggling to see words pop up below a person’s face when you’re standing in the real world and not in some digital environment.

Next, I approached several books on a shelf. I walked up and chose one. I opened the book to a random page and asked Gemini to summarize the page, and with shocking speed it did the deed and even highlighted a dialog between two characters. Another neat trick is that Gemini has a short window where it remembers things you see, so after I walked away, I was able to ask for the name of “the yellow book” on the shelf (which I did not even look at myself), and it quickly gave me the name. This context-aware window is “minutes long,” so it’s not like you can ask Gemini to remind you of something you saw early in the day, though Google says it’s working on increasing this window.

The demos continued. I walked up to a Nespresso machine and asked Gemini to tell me how to use it. (It got the model name right on the first try just from seeing it in front of me.) I looked at a record of OK Computer and asked Gemini to play a track from Radiohead’s album through the speakers on the glasses. Since Gemini 2.0 is multilingual, I also asked it to describe another painting in front of me in my second language, Malayalam. I’m not a native speaker, but I was able to pick out words and it sounded mostly there. It’s something I can imagine my mom taking advantage of, as English is not her native language.

Since it’s paired with your phone, your phone notifications will show up on the glasses as well. These are AI summarized so you don’t feel like a wall of text is shoved in front of your face, and you can use Gemini voice commands to respond. You can take pictures with a tap of a button on top of the glasses’ right arm, and a preview pops up on the glasses after the capture so you can see if you need to readjust your framing. Since there’s a display built in, these glasses can also play videos, and I watched a short clip. The screen was small, and it wasn’t the clearest picture, but Google says the final version will be several times brighter and more vivid.

Glass, a Decade Later

I remember watching the first-ever Google Glass demo in my college dorm room—truly an iconic moment at Google I/O 2012, where people skydived toward the Moscone Convention Center wearing cyborg smart glasses that were streaming video of their approach over a Hangouts call. These Android XR–powered smart glasses don’t command that much fanfare but, in my limited time with them, I can say this: Of all the smart glasses I’ve tried, they come the closest to realizing the original vision of Glass.

But Google is also in a very different place as a company than it was in 2012. A judge recently ruled Google Search to be an illegal monopoly, calling for the company to sell off Google Chrome. Yet Google (with Samsung) now wants to be the platform for the next wave of spatial computing. VR also has had a rocky road due to wavering consumer interest, and given Google’s history of killing off projects, it’s difficult to glean whether a face computing platform that requires special (and expensive) hardware will meet the fate of so many apps and services that came before.

Izadi says the platform approach helps in that regard: “I think once you’re established as an Android vertical, we’re not going away anytime soon, so that’s kind of a guarantee we can give.”

The big bet seems to be around Gemini and AI. Oh, and the synergy between Google and Samsung. As Kihwan Kim, the executive vice president at Samsung spearheading Project Moohan, says, “This is not about just some teams or company making this—this is different. It’s completely starting from the ground up, how AI can impact VR and AR.” He went on to say the collaboration with Google felt like “one single spirit,” adding that it’s something he’s never experienced before in this line of work.