GUI-CC: Benchmarking Contextual Consistency of GUI World Models as Agent Environments
Source
arxiv.org
Author
Lin Fu, Zheyuan Yang, Tianhui Zhang, Jinbiao Wei, Guo Gan, Boxu Liu, Yilun Zhao, Yu Rong
Date
Context
A simulated app screen can look convincing while forgetting what happened earlier in the task. GUI-CC evaluates that problem in models that generate interface states for software agents. One track follows recorded mobile interactions; another lets fixed agents act on generated screens, so each result becomes the starting point for the next action.
The authors report that usable-looking individual screens often fail to preserve task context windowThe maximum amount of text a model can consider at once — its working memory for the current conversation or task.Full definition → or sustain executable interaction sequences. Their benchmarkA standard public test set for comparing AI models — the shared scoreboards behind every "model X beats model Y" claim.Full definition → therefore tests whether a generated interface remains a coherent environment over time, which matters when using it to evaluate agents that must complete several connected actions.
Terms in this piece · Glossary
context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
benchmark — A standard public test set for comparing AI models — the shared scoreboards behind every "model X beats model Y" claim.
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
articleWebUIProof: Benchmarking WebUI Code Generators with UI-Agent Execution HarnessYun-Yun Tsai, Yuning Mao, Shiqi Wang, Junfeng Yang, Sinong Wang
articleUI-Venus-2 Technical ReportVenus Team, Zhuohan Cai, Haoxing Chen, Jiaxuan Chen, Weizhi Chen, Changlong Gao, Zhangxuan Gu, Yuan Guo, Yusong Hu, Jianrong Jiang, Jianguo Li, Runze Li, Jinzhen Lin, Zhenyu Ma, Changhua Meng, Han Peng, Xinyu Qiu, Shuheng Shen, Zhongyi Shui, Weiqiang Wang, Ming Wen, Zhuoer Xu, Hang Yan, Kaiwen Yang, Ruilin Yao, Nanjun Yu, Zhengwen Zeng, Lianrui Zhang, Yunzhu Zhang, Zhe Zhao, Beitong Zhou
articleDo GUI Agents Know When Not to Act? Enabling Conflict-Aware Termination for Multimodal GUI AgentsZhaoyuan Huang, Tianjie Ju, Pengzhou Cheng, Zheng Wu, Yansi Li, Chuanbiao Song, Jun Lan, Huijia Zhu, Weiqiang Wang, Zhuosheng Zhang