InSpatio

We don’t generate videos. 🎬 We generate worlds from videos. 🌍 Introducing InSpatio-World — the world's first open-source real-time 4D world model‼️ Your input: a video clip Our output: a dynamic, navigable, persistent world 🕹️ explore freely across viewpoints ⏪ control time forward and backward 🔓 open-source and ready to build on :) Live demo: 🔗 world.inspatio.com Code & weights: 🔗 github.com/inspatio/inspa… Project page: 🔗 inspatio.github.io/inspatio-world

Thank you @_akhaliq for sharing our work😉

@ChrisAtKIRI @oreki_galahad Quick update: we’ve released a faster, optimized version on GitHub. ⚡️ Also, our technical report is now available on arXiv for more details.

InSpatio-World (inspatio.github.io/inspatio-world/) looks so great, so our guy @oreki_galahad tested right away. Basically, the pipeline runs 3 steps: Use Florence-2 to generate video captions, estimate depth with DA3, and Run InSpatio-World v2v inference. Here is the result below. Amazing, but still has room for improvement in consistency. But generally I can dream a lot with it.

INSPATIO-WORLD technical report is now on arXiv. 🌎 A real-time 4D world model built for interactive spatial intelligence. From a single reference video, we reconstruct a dynamic world you can freely explore across viewpoints and time. tech report: arxiv.org/abs/2604.07209

@ChrisAtKIRI @oreki_galahad The current release of InSpatio‑World is an unoptimized version. The fully optimized version, consistent with our online live demo and reaching ~24 FPS on a datacenter GPU and 10 FPS on an RTX 4090, will be released together with our technical report within 2 weeks. Stay tuned!🪄

@ChrisAtKIRI @oreki_galahad This is awesome — thanks for diving in so fast 🙌 Great to see people already experimenting with different pipelines 😉🌐

@DLKFZWilliam2 Thanks for the love! We’re already seeing developers start building their own projects with it, so feel free to jump in 🙌 We’ll also be releasing a tech report soon. 🔜🌐

我，天啊，这个项目就是我一直想找的。 现在开源了，大家可以去看一看，能不能部署到自己的项目里面！ 项目名称叫做:InSpatio-World，是全球首款开源Apache-2.0 4D世界模型！ 支持任意视角切换 + 时间前进/倒退，我感觉稍微做一个 UI 就可以做一个很好玩的游戏😍

@ysu_ChatData Will share a full technical write-up very soon. Stay tuned! 👀🌐

@InSpatio_AI Nice launch. The real unlock for 4D world models is whether they stay useful after the demo: stable geometry, controllable cameras, and a clean way to edit world state over time. Curious how persistent the scene is once the video ends.

We don’t generate videos. 🎬 We generate worlds from videos. 🌍 Introducing InSpatio-World — the world's first open-source real-time 4D world model‼️ Your input: a video clip Our output: a dynamic, navigable, persistent world 🕹️ explore freely across viewpoints ⏪ control time forward and backward 🔓 open-source and ready to build on :) Live demo: 🔗 world.inspatio.com Code & weights: 🔗 github.com/inspatio/inspa… Project page: 🔗 inspatio.github.io/inspatio-world

@ByFrustrated Runs smoothly on a single 4090 in our tests 👍

@InSpatio_AI Do we know what you reqs are?

@JohnDoe_2333 4D rickroll coming 🔜🌐

@InSpatio_AI I guess it's time to make some 4D memes lol

@Elon114514_ That’s exactly the idea😉

@InSpatio_AI This crosses the line from media generation into environment generation

@mervenoyann Haha sorry for pulling you back in! We’re trying to keep things moving fast on our side too 🌏

it's raining 3D/world models I was going to take time off but just cancelled to catch up! 🥲

Our new 4D world model, InSpatio-World, is now open-sourced. It enables world generation directly from video inputs. 🌍

The technical report for InSpatio-WorldFM is now on arXiv. 📄 A frame-based world model enabling real-time interactive 3D world generation with multi-view spatial consistency. 🌍✨ Tech Report ↓ 🔗 arxiv.org/abs/2603.11911

Well… this blew up faster than we expected.🤯 InSpatio-WorldFM demo servers are currently overloaded. We’re scaling up ASAP. You can also run the open-source version locally — unlimited time and fully persistent👇 inspatio.github.io/worldfm

@jiqizhixin Thanks for the write-up! Glad to see people exploring what real-time world models can enable.😉

Wow, you could explore entire 3D worlds in real time using just a single consumer-grade GPU! @InSpatio_AI presents InSpatio-WorldFM, and it bridges the gap between static 2D images and interactive spatial intelligence. The model moves away from the simple pixel-guessing used in standard video AI. Instead, it uses a hybrid memory system with explicit 3D anchors that act like a digital blueprint. This ensures that the environment stays fixed and stable as you move through it, preventing objects from drifting or disappearing like they do in traditional video generation. By optimizing for efficiency, InSpatio-WorldFM achieves interactive frame rates on a single RTX 4090! It provides superior multi-view consistency compared to standard 2D diffusion models, making it a breakthrough for embodied robotics, real-time gaming, and instant 3D world building from a single photo.

@KeunhongP Haha, glad you noticed:) FrameBoy was definitely a major inspiration to us. Huge respect for the awesome work from World Labs!

Love the homage to the frameboy lol

Experience real-time 3D Al today. Try InSpatio-WorldFM directly on your device: 🔗 inspatio.com/worldfm 📱💻 We've open-sourced our weights and code. Build your own world model here: 🔗 github.com/inspatio/world… 🛠️ Project Page: inspatio.github.io/worldfm

Spatial Intelligence just went Open Source. 🌐 Introducing InSpatio-WorldFM — the world’s first open-source real-time interactive 3D world model. Last October, @theworldlabs's RTFM showed the vision. Today, we’re bringing that idea into the open for everyone. ✅Efficient, multi-view consistent spatial reasoning ✅Real-time interactive exploration ✅Open-source weights & code Stop watching demos. Start building now. 👇