arXiv 2025 · Preprint
PAN: A World Model for General, Interactable, and Long-Horizon World Simulation
Jiannan Xiang, Yi Gu, Zihan Liu, Zeyu Feng, Qiyue Gao, Yiyan Hu, Benhao Huang, Guangyi Liu, Yichi Yang, Kun Zhou, Davit Abrahamyan, Arif Ahmad, Ganesh Bannur, Junrong Chen, Kimi Chen, Mingkai Deng, Ruobing Han, Xinqi Huang, Haoqiang Kang, Zheqi Liu, Enze Ma, Hector Ren, Yashowardhan Shinde, Rohan Shingre, Ramsundar Tanikella, Kaiming Tao, Dequan Yang, Xinle Yu, Cong Zeng, Binglin Zhou, Zhengzhong Liu, Zhiting Hu, Eric P. Xing
Overview
PAN predicts future world states as videos conditioned on past observations and natural-language actions. Its Generative Latent Prediction architecture pairs an autoregressive latent dynamics model with a video diffusion decoder, connecting language-based reasoning with visual simulation. The work evaluates action-conditioned simulation, forecasting over long horizons, and reasoning through simulated futures.
Resources
This entry refers to the November 2025 version of the paper, which lists Enze Ma as a coauthor.