One of the Top 5 most valuable companies in the world has just acquired a license for Over the Reality 3D mapping dataset to train next-generation vision foundation models. This is a huge milestone for OVER 🧵👇

Before AI can understand the physical world, it needs data about the physical world. Not just images. Spatial environments. Real streets, buildings, parks, stores, museums, and public spaces captured in 3D. OVER has already reconstructed 290,000+ real-world 3D scenes, creating a growing dataset for the next generation of spatial and embodied AI. And this is exactly the kind of real-world spatial data AI companies are looking for. We're already in conversations with multiple AI and Robotics companies exploring how OVER's dataset can support the development and training of models that need to understand the physical world. The next AI race will be about real-world data.

And under the hood there’s a ton of cool stuff: VPS anchoring without prior scanning, new 3D asset generator, support for GSplat assets in the scene, and free test publishing on any OVRLand for 24 hours!

Build on top of the real world, directly from your phone. The OVER In-App Builder is now live. Open the app, create an experience, and place 3D assets directly into the physical world around you, in real time. And if the asset you need doesn't exist, you can create it on the spot with our AI 3D Asset Generator. Just describe what you want, generate it, and place it directly into your experience. Now available FOR FREE in the OVER app. link.ovr.ai/download

Imagine building a video game inside a real place. Capture a large environment with an Insta360 X5, walk through it, and let OVER reconstruct it as a large-scale Gaussian Splat. Now you have a digital version of that real-world space. Give it to AI and start building on top of it: add characters, missions, interactions, game mechanics, simulations, whatever you can imagine. Instead of creating the world from scratch, the real world becomes your starting point. And gaming is only one possibility. Once a place has been captured and transformed into reusable 3D spatial data, the same environment can become the foundation for XR, Robotics VPS, AI training, navigation, simulation and entirely new applications. Capture reality once. Build anything on top of it. And you can start doing it yourself with the Over the Reality app: link.ovr.ai/map2earn

Simulating a fly brain 🪰 inside of an @OVRtheReality 3D Map! What a time to be alive!!!

We’ve updated our Insta360 X5 Mapping Guide. 📍 One of the most important changes is simple: keep each mapping session under 10 minutes and return to where you started. Ideally, walk for around 5 minutes, turn around, and map your way back to the same starting point. Why? Because the way you capture a location directly affects the quality of the 3D reconstruction. Better capture paths mean better geometry, more reliable reconstructions, and higher-quality Gaussian Splats like the one you see here. If you’re mapping with an @insta360 X5, make sure to follow the updated guide before your next session. Read the updated Mapping Guide: overthereality.ai/Insta360X5… Own an Insta360 X5? Start mapping and earning with OVER: link.ovr.ai/map2earn Using another 360° camera? Generate your Gaussian Splat for free on @FreeGaussian, powered by OVER: freegaussian.ai/

A Gaussian Splat can become a world where Robots and AI agents can act. In our latest OVER Research experiment, we placed a robot inside a real-world 3D capture, with a VLM making decisions based on what it sees. At every step, the robot holds a pose in the reconstruction, gets a newly rendered view of the environment, takes an action, moves, and sees the world again from its new position. Why does this matter? Because 3D captures can become more than reconstructions to explore. They can become environments where embodied AI and robots can navigate, act, be evaluated and eventually train across real-world spaces at scale. Capture a place once. Then turn it into a world where AI and Robots can act. The full experiment, including what we discovered once we actually put the loop to the test: research.ovr.ai/posts/2026-0…

We're proud to announce our partnership with @insta360 By combining Insta360 cameras with OVER's Spatial AI technology, we're making it dramatically easier to capture large real-world environments and transform them into persistent digital twins. We already put this workflow to the test at Città Fiera, one of Italy's largest shopping centers. A single operator walked through the venue with one Insta360 X5 and captured 150,000 m² and 230+ stores in just 90 minutes. OVER then transformed the footage into a high-fidelity digital twin powering its VPS, enabling centimeter-accurate indoor localization and AR navigation directly from a smartphone browser. The same digital twin can then power much more than navigation: centimeter-accurate VPS localization, persistent XR experiences, remote virtual exploration, robotics localization, simulation environments, spatial computing applications, and training data for next-generation AI models. This is the value of the partnership: making the capture of the physical world simple enough to scale, while OVER turns that capture into spatial infrastructure that can support multiple products and use cases. Read the full story on the Insta360 website: insta360.com/blog/insta360-X…

Can 1,000 real-world 3D scenes make a state-of-the-art model better? At OVER Research, we tested it by fine-tuning R3, a state-of-the-art streaming 3D reconstruction model, on roughly 1,000 scenes from the OVER dataset. The result: 25% lower trajectory error on our real-world environments compared to the deployed public checkpoint. And we did it using only 1,000 scenes, the same open dataset anyone can download for free, out of the 285,000+ real-world 3D scenes OVER has today. Read the full OVER Research publication: research.ovr.ai/posts/2026-0… Download the 1,000-scene dataset on Hugging Face: huggingface.co/datasets/Over…

How accurate is our automatic 360° reconstruction pipeline? We tested it against a LIDAR survey captured at the same time, on the same rig, over a 118m path. The result: 3.9 cm median agreement with LIDAR, with no measurable accumulated drift across the full trajectory. The one systematic discrepancy we found was a uniform 2.9% scale offset. Our reconstruction reproduces its own metric reference to 0.03%, showing that the geometry itself did not create the error. In other words, the hard part, reconstructing shape consistently over a large scene, is already working at centimeter-level agreement. Read the full OVER Research publication: research.ovr.ai/posts/2026-0…

We mapped an entire shopping mall - around 50,000 m² - by walking it with an Insta360 X5. 17,959 fisheye camera poses, all registered. Then we relocalized an ordinary phone video against it, frame by frame. 99.9% of frames relocalized. Median pose jitter: 1.7 cm.