MolmoAct 2: An open foundation for robots that work in the real world

Abstract

MolmoAct2 is an open family of action reasoning models for robot control and real-world deployment. It builds on an embodied reasoning VLM backbone, integrates action modeling with a flow-matching action expert, and is released with open checkpoints and datasets for training, fine-tuning, and deployment.

Publication
arXiv preprint
Jaemin Cho
Jaemin Cho
Young Investigator @ AI2
Incoming Assistant Professor @ JHU

Incoming Asst. Prof. @ JHU working on Multimodal AI