I found myself facing a wall, talking to a smiling molecule that emerged from a vase. I had pinched my fingers over the very real vase, and now a very unreal molecule floated in front of me, googly eyes blinking, as it told me about its properties in ceramic material. This whole chat, AI-generated via Gemini, emerged as a hovering 3D graphic on glasses I was wearing. They were plugged into a phone-sized device running an app that was apparently coded on the fly in just days.

This advanced AR glasses setup, known as Project Aura, is just one of three Google glasses products coming this year. And there's more -- also on deck are wireless AI glasses made with Samsung, Gentle Monster and Warby Parker, slated for the fall, ready to compete with Meta's Ray-Ban and Oakley smart glasses. I spent time with all of them, and even a prototype dual-display set of wireless glasses, at Google's Mountain View campus. Back in December, I'd done a demo as well, but my latest deep dive showed me more, and better, things that all of these glasses can do. I was wowed.

Will Google help make smart glasses feel more everyday and useful? Or solve how offputting and potentially threatening their cameras and AI tech make some people feel? My latest set of demos at Google's campus, part of Google's AI-centric I/O developer conference, convinced me that they could be the best smart glasses out there. They could even help redefine what smart glasses can be. And they're the first glasses I've seen that really feel ready to work with the apps and services we have on phones already. 

In 2026 we've seen a flood of smart glasses, a category of product that most people don't wear and many are increasingly wary of. But Google's Gemini-powered glasses could change what's possible more than nearly any other glasses I've seen so far. The reason, clearly, is the AI and software under the hood -- and what Google's doing to lay groundwork for where Apple is reportedly heading too. That's even clearer to me now after meeting again with Google's head of XR, Shahram Izadi, and Samsung's Jay Kim, head of that company's MX customer experience division.

Google's glasses look ready to one-up Meta on AI smarts

I test-drove multiple versions of the smart glasses that Google and Samsung are releasing later this year. What I saw were still prototype models, but their more watch- and phone-connected features have more AI skills than Meta's latest glasses. But they still don't have prices yet, or even an official name.

Called "Intelligent Eyewear" for now, the glasses will be given product names later on by Warby Parker, Gentle Monster and Samsung when they're released this fall, according to Izadi, who spoke with me at Google's headquarters.

Watch this: The Future of Smart Glasses Is Coming This Fall

07:00

The glasses will all have cameras, microphones and embedded speakers like Meta's line of smart glasses, and also will come in display-free and single-display versions. The always-on Gemini Live mode on these glasses could follow me around a room and describe plants or tell me how to play a board game. I've seen these tricks before. But Google's advantage over Meta is how these AI tools relay back to apps already on phones, like Google Keep and Google Calendar. Or appear on watches. 

A translation demo I tried showed how automatic language recognition could identify French or Portuguese, spoken by people in the room with me. I saw on-screen captions in the display-enabled glasses, and audio translation relayed at a slight delay, tone-matched to speaking cadence and perceived speaker gender.

The glasses can also show AI-generated widgets that can work like mini apps to show quick-glance information like weather, scrollable in a stack.

Lightweight, in multiple designs

These glasses feel lighter-weight than Meta's Ray-Ban Displays: At 49 grams, they felt close to the heft of display-free Meta Ray-Bans. The two Warby Parker and Gentle Monster designs weren't something I got to try yet, but the prototype pair I wore is meant to represent a similar size and comfort, and they felt good despite having a single display inside one lens. They're definitely smaller than the equivalent, larger pair of display-enabled Ray-Ban Displays.

The Warby Parker and Gentle Monster designs look fun, though, at least in photos. Warby's rounded edges look good, with a frame that still seems somewhat thick. The oval-shaped Gentle Monster frames are definitely going for a different look, and I'm curious how much more other design types flex. There will be more, Izadi and Kim confirmed.

The prototypes I wore again, meanwhile, look more like Ray-Bans: generic and black-framed but functional.

Connecting with watches as well as phones

Photos I took on the glasses popped up moments later on a Pixel Watch worn by one of Google's team, showing off how a WearOS smartwatch could be a connected camera viewfinder. I tried taking a photo of the room with the glasses and adding Nano Banana effects: a David Lynch vibe and 1930s cartoon characters eating sushi. The image was updated in under a minute on the glasses, and it also popped up on the watch and the phone.

The glasses have a touchpad in one arm, which works with angle and double taps and swipes (single tap-hold for Gemini, double finger swipe for volume), and a camera shutter button on the top of one arm. They could also be controlled to some degree with a watch, although I didn't get to try that. Watch gestures would help, especially since Google doesn't have anything like Meta's neural band yet.

Dual displays are coming, too

Google's also working on dual-display glasses, which might not come until at least 2027. The prototypes I demoed again were notably heavier, and while I also tried this demo last year, this time the quality looked better for playing a Spider-Man clip converted to 3D by YouTube, and playing back 3D models that I could move with tilts and turns of my head, almost like augmented reality. Scrolling widgets moved by in 3D too.

The dual-display model is still being finessed, and Izadi and Kim acknowledged that there's no release timeframe yet. It's next on deck: maybe next year or the year after. But Izadi expects dual, single and no-display glasses to be on sale together down the road, likely at a range of prices.

AI privacy measures? Unknown

Samsung's and Google's teams admit that the world of wearables is a challenge to figure out and that public apprehension about face tech and camera-enabled glasses is something that still needs to be figured out. Meta's already faced multiple privacy problems and has muddy lines on its AI privacy policies. Google plans to go into more details on data privacy on its glasses at a fall event, says Izadi. 

"We've been doing a lot of thinking from the very beginning with these kinds of devices. Given the privacy aspects, you have to design with privacy in mind from the very, very beginning and leverage some of the standards that have been set in the AI world," says Izadi, who adds that a bystander LED indicator, fraud detection for AI and other features are part of the glasses. 

But expect the privacy rules for Gemini to work much like Gemini already does on phones, although glasses will have more situations where they could run in a live, always-on mode. Izadi says some of the challenge is "user education" but that when it comes to privacy "we do need to raise the bar, for sure."

"I think market expectation is higher on us," Samsung's Kim adds. "We're studying a lot. We've been thinking about it for a long time, from the beginning."

Prescriptions shouldn't be a problem

I did my demos of all these glasses using custom inserts matched to my heavily myopic prescription, but Google and Samsung are confident that their range of all-day smart glasses with Warby Parker and Gentle Monster will support prescription ranges including my own, even in display-enabled models. 

"We are really going deep there," says Izadi. "This is where some of the partnerships with the eyewear companies have been really important." Izadi points to the display-enabled Even Realities G2 glasses that I've worn using my prescription as "existence proof" that they can be made.

I hope so, because I plan to test out Google's glasses as my everyday pair. Unlike Meta's prescription support-limited Ray-Ban Displays, it looks like that could actually happen.

Google Health? Possible at some point

Google's new Fitbit Air and Gemini-powered health coach, which I'm test driving now, already had me thinking about how fitness and health could be a part of the next wave of glasses. Izadi acknowledges that the glasses could access data like this and interconnect, just not right away.

Kim points to the triangulation between rings and watches and glasses and phone. 

"Everything will come together in a way to support each other," he says, but also adds "You don't want to overwhelm the consumer, it has to be very natural and fit into your body." 

At its Unpacked event in July, Samsung is going to get into more glasses details and reveal more Gentle Monster and Warby Parker designs, so we may learn more on the fitness front there. But it's clear that Samsung and Google are going to make it happen. Meta's fitness focus with its glasses and Garmin, and the whole outdoors lifestyle pitch of glasses, makes it an obvious fit.

Project Aura: A pocket-size Gemini computer plus glasses

Now let's get back to that talking molecule demo that stunned me, which came courtesy of Google and Xreal's Project Aura, one of the most fascinating products I've seen in a while. It's not the first time I've tried it, but as Aura readies to launch later this year, this new batch of demos reminded me what amazed me back in December. In many ways, this glasses-plus-processor-puck combo delivers what a VR headset can do, and possibly more, in a little package that also works with your phone, laptop or Steam Deck.

It doesn't work with your phone directly. Instead, it has a processor puck that runs Android XR and Android apps, relaying this to glasses. It's very much a living model of what Meta's Project Orion prototype AR glasses were going for two years ago when I tried them, but with hardware available now, and with hand tracking via cameras instead of using a gestural neural band like Meta has. 

Anyone who's ever tried a pair of display-enabled tethered glasses, like those that Xreal or Viture or TCL sell, might know how good these "headphones for your eyes" can look. Project Aura is better, extending the field of view of its microOLED displays to 70 degrees, effectively enough to fill the glasses and not feel like you're missing too much.

Aura, as I said last year, can do all the things Samsung's 6-month-old Galaxy XR mixed reality headset can do -- but in a smaller package. The phone-sized processor puck has a touchpad surface and fingerprint ID button, and runs all the Android apps and VR/AR-enabled experiences that Galaxy XR does. The glasses have cameras and depth sensors that can enable hand tracking, pinching and scrolling apps and playing VR games. 

But this little package can also run a lot of AI-intensive things. Gemini Live can run in conjunction with any app, but you can also plug in other devices via USB-C/DisplayPort. Connecting a Steam Deck meant I could play Hollow Knight on the glasses on the big screen, but also open a YouTube window and drag it next to the game in any position, watching run-throughs. It can support five apps open at once, or have Gemini Live watch my Steam Deck gameplay and give playthrough advice. I could connect an iPhone or a MacBook or any other device and layer it with Gemini Live analysis, too.

Aura as a road to AI coding for XR?

What really wowed me was seeing an app vibe-coded with Google tools that ran right in Aura via WebXR, demonstrating how Aura could potentially be used to AI code and preview 3D experiences that I've never seen done before. One app I tried, a 3D drawing experience called 3D Paint -- I used my fingers to sketch in the air like the old VR app Tilt Brush -- was was vibe-coded in Chrome on Project Aura using Gemini Canvas. Apparently it took minutes to do.

Two other apps -- a Gemini music experience called Gemini Melody where I sketched lines in the air and had Gemini turn them into musical tones and interactive music tracks, and that talking molecule experience, called Gemini Molecule -- were standalone XR apps made in a week, coded using Google Antigravity and Gemini APIs. They're all demos of what sorts of AI-powered XR apps could be made and tested with Project Aura. If this is as easy to do as Google makes it seem, Project Aura could transform the VR/AR landscape in wild ways. It makes me want to try to vibe code.

Aura shows where glasses and phones could head next

Aura will go on sale as an actual product later this year, but at an unknown price. I expect it might be a more expensive product than I'd like, but it's also truly a shrunken-down set of some of the same pieces that Vision Pro and Galaxy XR were aiming for, in a form I'm more likely to use on the go. It's part dev kit, part portable VR replacement and preview of where AR glasses that connect with our phones could eventually evolve.

Imagine the phone-sized processing puck that Aura uses being the phone in our pockets instead. Current phones can't do it, but Aura feels like a preview of what's possible. 

"No plans yet, but it's an obvious next step, I would say a logical next step," Izadi says to me of taking the Aura experience into glasses that work with our phones directly. But that might take some time. 

Kim adds, "I think the phone will probably stay as a sort of central hub for many years to come," noting that phones and watches are destined to stay connected to phones down the road. 

I agree. But it's clear that the capabilities of those connections are just getting started. Project Aura is going to arrive sometime later this year alongside the wireless Warby Parker and Gentle Monster glasses, but it's a more advanced preview of a full computer with Gemini… and, really, the AR glasses I've been waiting for.