If you're building with World models. Apply for our Creative/Startup program for a chance to get free credits, early access and direct team feedback.

High-quality 3D reconstruction is still hard. In particular, with 3DGS, getting full coverage without occlusions is often impractical: training views can look great, but you can barely move around. With Echo-2, we take a different path: reconstruction as conditional generation. By leveraging a strong generative prior, Echo-2 creates high-fidelity digital twins from sparse input views—constrained by what was captured, while completing appearance and geometry in a coherent 3D world. These are a few sample scenes, but the same approach scales to large environments from only a handful of images. API update incoming!

Sparse captures in, complete worlds out. Echo reconstructs scenes from a small set of views, using world modeling priors to fill what sparse 3DGS leaves behind. Coming soon to the SpAItial API.

This is the easiest way to use the 3D world model from @SpAItial_AI. I asked Claude to recreate worlds from my childhood photos, and now I can explore them and walk around inside. The nostalgia is incredible 🤯 You can just ask for edits, generate collision meshes, and fully vibecode environment and games using Claude or any agent

Today, we enable AutoResearch in the physical world for the first time! Introducing ENPIRE: we give 8 Codex agents a fleet of robots, an allocation of GPUs, and generous token budget. We set them free with a simple goal: solve the task as quickly as possible, keep the robots busy but stay safe, don't waste precious compute. Make no mistake. Then humans step aside and our watch begins. The robot fleet starts to come alive: they learn to look for visual clues, reset the scene, practice novel skills, tinker with control stack, read papers online, debate, reflect, get stuck, and try again directly on the hardware. All we did is to give Codex an API to the world of atoms, and the rest is emergence. ENPIRE is able to solve high-precision tasks like tying zip-ties, organizing fine pins, and installing GPUs all by itself. We also discovered a new type of "physical scaling": 8 robots exploring in parallel improves significantly faster than fewer ones. A part of our NVIDIA GEAR lab now self-improves tirelessly over night. We just read the reports in the morning. /goal: we all take a holiday and Jensen wouldn't even notice ;) We will be open-sourcing everything, so you can host your self-running robot lab at home too! Deep dive in the thread:

Added some new scenes to show the bot using learned navigation skills. Navigation on Planet Mars Navigation on a skate park scene Works really well, and avoids new obstacles i throw at it. It also learned to jump! 🤯

Photoreal walls. Real geometry. Bots. WebRTC and all open source. Here's how built a fully multiplayer, browser-playable Quake III arena where every level is a 3D world created from a single Image using Echo-2 🧵👇

Having a few hours this afternoon after @CVPR? Consider visiting Red Rocks 20 minutes outside of Denver! For some inspiration, here's a world processed by @SpAItial_AI Echo-2 HQ model from a single phone image! Link to the world in🧵

❄️ New @playcanvas engine example: real-time Depth of Field on Gaussian Splats, powered by the engine's post-processing stack — with scene depth driven by a proxy mesh. Thanks to @SpAItial_AI for the Gaussian Splat mesh generation 🙏 Live link 👇

Excited for my workshop talks @CVPR in Denver! I'll cover recent works from our PhD students on generative AI and physically-grounded world models such as GenRecon, World Reconstructions, GaussianGPT, and many more. Bonus: some of our latest results from @SpAItial_AI :)

🚀 Major upgrades just landed in SuperSplat, the free and open source platform for 3D Gaussian splatting. Here is a 24-MILLION-Gaussian scan streaming live to a web browser. Near instant load time. Solid 60 fps. How? A new compute-based WebGPU renderer automatic LOD streaming. 🧵

I am in the back row of the CVPR AC workshop RIGHT NOW fragging noobs inside a radiance field, username is "Jon Barron", come at me, I dare you.

We turned 3D worlds created with Echo-2 into a multiplayer, Quake3, combat game. Imagine a world you generate from a single image becomes a full playable game map. 🔥 Play live now 👇🏽

Enter the Backrooms - in 15 minutes 🪄 Upload a movie still to @SpAItial_AI Echo-2 ⏬ Download a 3D Gaussian splat PLY file 🥽 Upload to @PlayCanvas SuperSplat and enter VR [1 / 3]

Releasing Echo-2 HQ, an improved model that delivers greater detail and sharper results. You can zoom in super close and discover remarkable appearance fidelity. Available via API and in the app. Try it out! Check out some of the scenes below👇