Humanoid Foundation Models

The ultimate resource for foundation models for humanoid robots and embodied AI. These models are also called Large Behavior Models. Here you can stay updated on the available humanoid foundation models, comparisons, benchmarking, ratings, specifications and more. And if you’re building a model the world should know about, we’d love to feature it. Feel free to use “Add your model” button at the bottom of this page.

How the benchmark works Each foundation model is independently rated on a 10-point scale across 10 capability dimensions: locomotion, whole-body control, bimanual coordination, dexterous manipulation, navigation, reasoning, sim-to-real transfer, cross-embodiment, real-time inference, and long-horizon planning. Use the filters below to narrow by model category – VLA, world model, or reward model – or by features like open weights, on-device inference, and humanoid-only support.