titleDual Exam: Controlled Evaluation of Multimodal Agents and World Models
authorsZhefan Rao, Meng Chu, Xuan Zhang, Jingfan Chen, Guohao Cai, Runyi Li, Shiyi Du, Ning Lu, Xu Huang, Zhixuan Chen, Feng Nie, Harold Haodong Chen, Hanwen Wang, Wen-Jie Shu, Yajing Bai, Xianfeng Wu, Xuanhua He, Shuhan Guo, Ziyue Ma, Ruichuan An, Long Bai, Miaosen Chai, Wenhao Chai, Jingye Chen, Qi Chen, Xuhang Chen, Wei Chow, Minghao Fu, Yunhao Gou, Haokun Gui, Dong Huang, Wei Huang, Zi Cheng Huang, Haibo Jin, Lingdong Kong, Leung Chun Ho, Tianqin Li, Yanheng Li, Jiayuan Liu, Jiyang Liu, Mingyu Liu, Shaoteng Liu, Xiaocheng Lu, Yawen Luo, Zeyu Qin, Tiancheng Shen, Yueqi Song, Zhaolong Su, Hengkai Tan, Tianhao Tang, Yiran Tao, Juncheng Wang, Qinghe Wang, Shangshang Wang, Yingying Wang, Fang Wu, Tiange Xiang, Qiushi Yang, Weicai Ye, Weichen Yu, Jiarui Zhang, Jize Zhang, Liuzhou Zhang, Renrui Zhang, Sheng Zhang, Wenhu Zhang, Junru Zhao, Lingting Zhu, Yuxin Zuo, Yutong Bai, Long Chen, Yang Deng, Ziqi Huang, Yeying Jin, Weijia Li, Kevin Qinghong Lin, Quande Liu, See-Kiong Ng, Hongliang Ren, Hongru Wang, Shujun Wang, Qifeng Chen, Haoxuan Che
date2026-10-22
venuearXiv (preprint, not peer reviewed)
idarXiv:2610.31066
upvotes34 (Hugging Face Daily Papers, 2026-11-01)
linksabs pdf hf
primerWorld models
abstractWorld Models (WMs) can support Multimodal Agents (MAs) through simulated experience, while Agent trajectories can inform World Model refinement. Evaluating this interaction requires separate evidence about action selection and state evolution. We introduce Dual Exam, which tests Agents under fixed environment response mechanisms and World Models under fixed action interventions across physical, digital, social, and scientific domains. The Agent track compares 10 configurations on 1,088 items in 15 reporting categories. The World Model track contains 400 cases across eight scenarios; 15 configurations have full coverage and one covers six scenarios. GPT 6 Astra leads the case-weighted Agent Index at 31.3, and GPT-Image-2 leads the full-coverage WM Overall at 21.7. These scores use different tasks and metrics and are not directly comparable. A 150-case pilot of one coupled pairing records zero task successes and motivates component-level diagnosis. Results reveal uneven task performance and failures to preserve task-relevant state. Their value for predicting coupled success or sustained reciprocal learning remains to be established.

cite

@misc{rao2026dual,
  title={Dual Exam: Controlled Evaluation of Multimodal Agents and World Models},
  author={Zhefan Rao and Meng Chu and Xuan Zhang and Jingfan Chen and Guohao Cai and Runyi Li and Shiyi Du and Ning Lu and Xu Huang and Zhixuan Chen and Feng Nie and Harold Haodong Chen and Hanwen Wang and Wen-Jie Shu and Yajing Bai and Xianfeng Wu and Xuanhua He and Shuhan Guo and Ziyue Ma and Ruichuan An and Long Bai and Miaosen Chai and Wenhao Chai and Jingye Chen and Qi Chen and Xuhang Chen and Wei Chow and Minghao Fu and Yunhao Gou and Haokun Gui and Dong Huang and Wei Huang and Zi Cheng Huang and Haibo Jin and Lingdong Kong and Leung Chun Ho and Tianqin Li and Yanheng Li and Jiayuan Liu and Jiyang Liu and Mingyu Liu and Shaoteng Liu and Xiaocheng Lu and Yawen Luo and Zeyu Qin and Tiancheng Shen and Yueqi Song and Zhaolong Su and Hengkai Tan and Tianhao Tang and Yiran Tao and Juncheng Wang and Qinghe Wang and Shangshang Wang and Yingying Wang and Fang Wu and Tiange Xiang and Qiushi Yang and Weicai Ye and Weichen Yu and Jiarui Zhang and Jize Zhang and Liuzhou Zhang and Renrui Zhang and Sheng Zhang and Wenhu Zhang and Junru Zhao and Lingting Zhu and Yuxin Zuo and Yutong Bai and Long Chen and Yang Deng and Ziqi Huang and Yeying Jin and Weijia Li and Kevin Qinghong Lin and Quande Liu and See-Kiong Ng and Hongliang Ren and Hongru Wang and Shujun Wang and Qifeng Chen and Haoxuan Che},
  year={2026},
  eprint={2610.31066},
  archivePrefix={arXiv},
  url={https://arxiv.org/abs/2610.31066}
}