RewardAI unveils OM-1 robot foundation model trained purely on human manipulation data

zipengfu · x · 2026-09-15

RewardAI introduces OM-1, its first robot foundation model that zero-shot generalizes to tabletop arms, industrial arms, and humanoids. Trained directly from human manipulation data with no teleop or robot data, it reaches near human-level dexterity and efficiency and supports multi-robot collaboration. OpenAI researcher npew highlighted that the progress comes from human manipulation, not teleop.

Related event: Reward AI Unveils Omnibody Stack and Robot Foundation Model OM-1(29 posts)→

Original post →

More from Embodied

Embodied channel →