Samsung has taken the lead as the first official partner to announce a mixed reality (MR) headset under the new Android XR platform. Dubbed “Project Moohan,” this futuristic device is set to hit the consumer market in 2025, and I’ve had the unique opportunity to test an early version.
Unfortunately, Samsung and Google are keeping the finer details—like resolution, weight, field-of-view, and price—tightly under wraps for now. Photos and videos were off-limits during my demo, but we do have an official image to tide us over.
Picture Project Moohan as a crossover between the Quest and Vision Pro, and you’ve got a good sense of its capabilities. It’s not just a vague comparison, either. The design heavily nods to Vision Pro, from its color scheme to the button layout and even the setup process, clearly showing its awareness of current market offerings.
Software-wise, imagine if Horizon OS and VisionOS combined forces, and you’d end up with something like Android XR. It’s remarkable how Project Moohan and Android XR mirror the functionality of these major platforms.
Let’s be clear though; this isn’t about copying, but rather about the tech industry’s norm of borrowing and refining good ideas. If Samsung’s new offerings capture the best features and sidestep the pitfalls, it’s a win for developers and consumers alike. It seems they’ve done just that, as many appealing features are present.
In terms of appearance, Project Moohan sports a sleek, goggle-like design reminiscent of Vision Pro. Unlike Vision Pro’s soft strap, which I find uncomfortable unless modified, Samsung opted for a firm strap with a tightening dial, akin to the Quest Pro’s ergonomic design. This open-peripheral approach makes it excellent for augmented reality (AR), and like the Quest Pro, it also accommodates snap-on blinders for when you want a more immersive experience.
While there’s a stark resemblance to Vision Pro in terms of button placement and overall structure, Project Moohan lacks the external display feature that shows the wearer’s eyes. Vision Pro’s ‘EyeSight’ display, although somewhat controversial, is a feature I wish Project Moohan had embraced, as it feels odd not seeing the person behind the headset.
Samsung is tight-lipped about Project Moohan’s technical specs, citing its prototype status. However, it is confirmed that it runs on the Snapdragon XR2+ Gen 2 processor, a more potent iteration of the chip found in the Quest 3 and Quest 3S.
During my time with the headset, I noted a few key details. It boasts pancake lenses with automatic IPD adjustment, thanks to integrated eye-tracking technology. The field-of-view seems narrower than that of the Quest 3 or Vision Pro, but only further testing with different forehead pads—which reportedly allow you to move closer to the lenses—will confirm this.
The field-of-view felt more compact, yet still quite immersive, though there was some brightness drop-off at the edges of the display. Again, this could improve if lenses are closer to the eyes. For now, I’d say Meta’s Quest 3 leads in lens technology, with Vision Pro following and Project Moohan after that.
Samsung confirmed Project Moohan would have separate controllers, but I didn’t get to experience them. They might ship with the headset or be available separately—the decision hasn’t been made.
The hands-on experience relied on hand and eye-tracking inputs, blending Horizon OS and VisionOS features. You can use raycast cursors like Horizon OS or eye+pinch inputs similar to VisionOS. It even has downward-facing cameras to track comfortable lap-positioned pinches.
Once I donned the headset, the immediate sharpness of my hands stood out. Compared to Quest 3’s cameras, Project Moohan’s passthrough cameras seemed to offer a crisper image, with less motion blur than Vision Pro under optimal lighting. It was as if these cameras were specially attuned to arm’s-length distances.
Delving into Android XR, it mirrors a fusion of Horizon OS and VisionOS. The home screen resembles Vision Pro, with app icons floating on a transparent canvas. Selecting an app opens floating panels via a simple look and pinch gesture, sharing much with Vision Pro’s interactions.
System windows lean more towards Horizon OS aesthetics, with largely opaque backgrounds and movable windows through an invisible framing system.
Not just limited to flat applications, Android XR can handle fully immersive experiences. A VR iteration of Google Maps felt much like Google Earth VR, allowing exploration of the globe, 3D cityscapes, Street View images, and newly volumetric interior captures.
While Street View remains monoscopic 360 imagery, volumetric captures are renderings that you can wander through in real-time. Google explained this as a gaussian splat solution—the specifics of whether this leverages existing Google Maps photography or requires new scans are unclear. Though not as crisp as photogrammetry might offer, Google assured me this was on-device processing and quality would improve.
Google Photos sees enhancements with Android XR, offering automatic 3D conversions of existing 2D photos and videos. From what little I experienced, these looked impressive—matching the Vision Pro’s quality.
YouTube also maximizes Android XR’s capabilities. Apart from enjoying typical flat content on a panoramic screen, users can delve into 180, 360, and 3D videos from the platform’s library. While not all are high-definition, their presence underscores a commitment to this medium—likely to grow as more headsets support it.
I checked out a YouTube video that had been auto-converted from 2D to 3D for headset viewing. It looked quite satisfactory, reminiscent of Google Photos’ quality transitions. It remains uncertain whether content creators have to opt into such conversions or if this will be an automatic feature—something we’ll learn more about over time.
In terms of standout features, Android XR and Project Moohan excel in conversational AI. Google’s Gemini AI, particularly the ‘Project Astra’ model, can be initiated from the home screen and offers a seamless, continuous understanding of both spoken words and the visual environment—both real and virtual.
While Vision Pro relies on Siri, which is audio-only and task-centric, and Quest’s Meta AI hears and sees only the real world, creating a gap when virtual elements are missed, Gemini gets closer to offering a cohesive experience with a video feed of your dual environments, eliminating pauses.
Gemini’s contextual memory is another leap. Retaining a rolling 10-minute memory of conversations and visual interactions, it brings continuity to your interactions. During a demo, querying objects in a room showed how adept it was at contextual recognition and responses.
I tested Gemini for translating a Spanish sign into English, then teasingly requested a similar French translation for another sign already in French. Gemini smoothly navigated this, even providing pronunciation with a French accent.
Later on, after a pause in our conversation, I enquired about sign translations—and Gemini impressively recalled the specifics, showcasing its memory and contextual prowess.
Asked to control the headset, Gemini demonstrated how integrated it is. Requesting “take me to the Eiffel Tower” produced an immersive 3D Google Maps view, allowing further interesting questions about the iconic landmark.
Gemini also leveraged YouTube by fetching relevant clips, for instance, showing a ground-level view when prompted—a testament to its clever integration of available information.
It’s safe to expect Gemini on Android XR to handle normal AI tasks—like sending messages or setting reminders—but its depth in XR-specific tasks remains to be fully revealed.
Gemini on Project Moohan currently stands as a leading headset AI, painting Apple’s and Meta’s existing capabilities in a trailing light. Nevertheless, both competitors are surely developing similar capacities. How long Google can maintain this lead is a question only time will answer.
While Gemini on Project Moohan offers a delightful value-add for productivity-focused use, its true potential could unfold on smaller, everyday smartglasses—an experience I also explored but will delve into in another piece.
