Diffusion foundation model for bimanual skill, trained with AgiBot and public data.
Leading open bimanual diffusion policy
RDT-1B (Robotics Diffusion Transformer) is a diffusion-based generalist policy around the 1B scale, aiming at scalable imitation across tasks and embodiments. It sits between compact policies like Octo and larger VLM-style VLAs.
Consider RDT when you want diffusion action modeling with more capacity than classic DP, while remaining in an open research release format.
Researchers studying diffusion generalist policies at ~1B scale.
Physical Intelligence generalist VLA (~3B) with flow-matching actions; π0 weights are partially open — strong dexterous-manipulation reference in the π0 family.
Content reviewed 2026-07-23
