We introduce RoboBrain 2.5, a next-generation embodied AI foundation model that advances general perception, spatial reasoning, and temporal modeling through extensive training on high-quality spatiotemporal supervision. Building upon its predecessor, RoboBrain 2.5 introduces two major capability upgrades. Specifically, it unlocks Precise 3D Spatial Reasoning by shifting from 2D pixel-relative grounding to depth-aware coordinate prediction and absolute metric constraint comprehension, generating complete 3D manipulation traces as ordered keypoint sequences under physical constraints. Complementing this spatial precision, the model establishes Dense Temporal Value Estimation that provides dense, step-aware progress prediction and execution state understanding across varying viewpoints, producing stable feedback signals for downstream learning. Together, these upgrades extend the framework toward more physically grounded and execution-aware embodied intelligence for complex, fine-grained manipulation. The code and checkpoints are available at project website: https://superrobobrain.github.io
Cite
@article{arxiv.2601.14352,
title = {RoboBrain 2.5: Depth in Sight, Time in Mind},
author = {Huajie Tan and Enshen Zhou and Zhiyu Li and Yijie Xu and Yuheng Ji and Xiansheng Chen and Cheng Chi and Pengwei Wang and Huizhu Jia and Yulong Ao and Mingyu Cao and Sixiang Chen and Zhe Li and Mengzhen Liu and Zixiao Wang and Shanyu Rong and Yaoxu Lyu and Zhongxia Zhao and Peterson Co and Yibo Li and Yi Han and Shaoxuan Xie and Guocai Yao and Songjing Wang and Leiduo Zhang and Xi Yang and Yance Jiao and Donghai Shi and Kunchang Xie and Shaokai Nie and Chunlei Men and Yonghua Lin and Zhongyuan Wang and Tiejun Huang and Shanghang Zhang},
journal= {arXiv preprint arXiv:2601.14352},
year = {2026}
}