Introducing MolmoSpaces, a large-scale, fully open platform + benchmark for embodied AI research. 🤖 230k+ indoor scenes, 130k+ object models, & 42M annotated robotic grasps—all in one ecosystem.

i have been using procthor to great effect. thank you for your service

The next wave of AI will act in the physical world, but building robots that generalize across new environments – rather than replaying learned behaviors – requires far more diverse training data than exists today. That's where MolmoSpaces comes in.

MolmoSpaces builds on two foundations: Objaverse, one of the largest open collections of 3D objects, and our THOR family of interactive simulation environments, all unified with physics-grounded simulation + validated physical parameters tuned for realistic manipulation. ⚙️

MolmoSpaces ships with a massive bank of validated grasps for rigid and articulated objects, loadable directly into environments. An accompanying trajectory-generation pipeline supports reproducible demonstrations and imitation learning at scale.

📐 MolmoSpaces-Bench is our new benchmark for evaluating generalist policies under systematic, controlled variation. Researchers can vary one factor at a time – from object properties to layouts, dynamics, and instruction phrasing – across thousands of realistic scenes.

MolmoSpaces supports teleoperation via mobile platforms like Teledex—collect demonstrations right from your phone, compatible with all our embodiment setups, including DROID and CAP; no extra configuration is needed.

All MolmoSpaces assets, scenes, & tools are open + modular, provided in MJCF with USD conversion for cross-simulator portability. Plug in new embodiments, regenerate grasps, & run across MuJoCo, ManiSkill, & NVIDIA Isaac Lab/Sim.

Explore MolmoSpaces & start building: 📝 Blog: 💻 Demo: ⬇️ Code: 📊 Data: ✍️ Paper:

This is awesome work! Curious—any plans to integrate SceneSmith-like agentic scene generation into MolmoSpaces? It feels like a natural combo: MolmoSpaces benchmark + SceneSmith prompt-to-sim scenes = infinite evaluation distribution.

凄いですね！めちゃくちゃロボット研究が進みそう！広めさせていただきました！

MolmoSpaces just turned embodied AI into an absolute playground

AHHHHHH

I can imagine how this paired with some skinning can be an absolute goldmine for RL.

free? sure, until they find a way to monetize your data. local means nothing when they own the code.

This is cool research but we have the best ai2 domain!