MolmoAct2: Action Reasoning Models for Real-world Deployment
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , , , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866913100746194944 |
|---|---|
| author | Fang, Haoquan Duan, Jiafei Clay, Donovan Wang, Sam Liu, Shuo Huang, Weikai Fan, Xiang Tsai, Wei-Chuan Chen, Shirui Wang, Yi Ru Xing, Shanli Cho, Jaemin Park, Jae Sung Eftekhar, Ainaz Sushko, Peter Farley, Karen Wadhwa, Angad Harrison, Cole Han, Winson Lee, Ying-Chun VanderBilt, Eli Hendrix, Rose Ellawela, Suveen Ngoo, Lucas Chai, Joyce Ren, Zhongzheng Farhadi, Ali Fox, Dieter Krishna, Ranjay |
| author_facet | Fang, Haoquan Duan, Jiafei Clay, Donovan Wang, Sam Liu, Shuo Huang, Weikai Fan, Xiang Tsai, Wei-Chuan Chen, Shirui Wang, Yi Ru Xing, Shanli Cho, Jaemin Park, Jae Sung Eftekhar, Ainaz Sushko, Peter Farley, Karen Wadhwa, Angad Harrison, Cole Han, Winson Lee, Ying-Chun VanderBilt, Eli Hendrix, Rose Ellawela, Suveen Ngoo, Lucas Chai, Joyce Ren, Zhongzheng Farhadi, Ali Fox, Dieter Krishna, Ranjay |
| contents | Vision-Language-Action (VLA) models aim to provide a single generalist controller for robots, but today's systems fall short on the criteria that matter for real-world deployment. Frontier models are closed, open-weight alternatives are tied to expensive hardware, reasoning-augmented policies pay prohibitive latency for their grounding, and fine-tuned success rates remain below the threshold for dependable use. We present MolmoAct2, a fully open action reasoning model built for practical deployment, advancing its predecessor along five axes. We introduce MolmoER, a VLM backbone specialized for spatial and embodied reasoning, trained on a 3.3M-sample corpus with a specialize-then-rehearse recipe. We release three new datasets spanning low-to-medium cost platforms, including MolmoAct2-BimanualYAM, 720 hours of teleoperated bimanual trajectories that constitute the largest open bimanual dataset to date, together with quality-filtered Franka (DROID) and SO100/101 subsets. We provide OpenFAST, an open-weight, open-data action tokenizer trained on millions of trajectories across five embodiments. We redesign the architecture to graft a flow-matching continuous-action expert onto a discrete-token VLM via per-layer KV-cache conditioning. Finally, we propose MolmoThink, an adaptive-depth reasoning variant that re-predicts depth tokens only for scene regions that change between timesteps, retaining geometric grounding at a fraction of prior latency. In the most extensive empirical study of any open VLA to date, spanning 7 simulation and real-world benchmarks, MolmoAct2 outperforms strong baselines including Pi-05, while MolmoER surpasses GPT-5 and Gemini Robotics ER-1.5 across 13 embodied-reasoning benchmarks. We release model weights, training code, and complete training data. Project page: https://allenai.org/blog/molmoact2 |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_02881 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | MolmoAct2: Action Reasoning Models for Real-world Deployment Fang, Haoquan Duan, Jiafei Clay, Donovan Wang, Sam Liu, Shuo Huang, Weikai Fan, Xiang Tsai, Wei-Chuan Chen, Shirui Wang, Yi Ru Xing, Shanli Cho, Jaemin Park, Jae Sung Eftekhar, Ainaz Sushko, Peter Farley, Karen Wadhwa, Angad Harrison, Cole Han, Winson Lee, Ying-Chun VanderBilt, Eli Hendrix, Rose Ellawela, Suveen Ngoo, Lucas Chai, Joyce Ren, Zhongzheng Farhadi, Ali Fox, Dieter Krishna, Ranjay Robotics Vision-Language-Action (VLA) models aim to provide a single generalist controller for robots, but today's systems fall short on the criteria that matter for real-world deployment. Frontier models are closed, open-weight alternatives are tied to expensive hardware, reasoning-augmented policies pay prohibitive latency for their grounding, and fine-tuned success rates remain below the threshold for dependable use. We present MolmoAct2, a fully open action reasoning model built for practical deployment, advancing its predecessor along five axes. We introduce MolmoER, a VLM backbone specialized for spatial and embodied reasoning, trained on a 3.3M-sample corpus with a specialize-then-rehearse recipe. We release three new datasets spanning low-to-medium cost platforms, including MolmoAct2-BimanualYAM, 720 hours of teleoperated bimanual trajectories that constitute the largest open bimanual dataset to date, together with quality-filtered Franka (DROID) and SO100/101 subsets. We provide OpenFAST, an open-weight, open-data action tokenizer trained on millions of trajectories across five embodiments. We redesign the architecture to graft a flow-matching continuous-action expert onto a discrete-token VLM via per-layer KV-cache conditioning. Finally, we propose MolmoThink, an adaptive-depth reasoning variant that re-predicts depth tokens only for scene regions that change between timesteps, retaining geometric grounding at a fraction of prior latency. In the most extensive empirical study of any open VLA to date, spanning 7 simulation and real-world benchmarks, MolmoAct2 outperforms strong baselines including Pi-05, while MolmoER surpasses GPT-5 and Gemini Robotics ER-1.5 across 13 embodied-reasoning benchmarks. We release model weights, training code, and complete training data. Project page: https://allenai.org/blog/molmoact2 |
| title | MolmoAct2: Action Reasoning Models for Real-world Deployment |
| topic | Robotics |
| url | https://arxiv.org/abs/2605.02881 |