GenExam: A Multidisciplinary Text-to-Image Exam

Zhaokai Wang, Penghao Yin, Xiangyu Zhao, Changyao Tian, Yu Qiao, Wenhai Wang, Jifeng Dai, Gen Luo · 2025

Read GenExam: A Multidisciplinary Text-to-Image Exam by Zhaokai Wang, Penghao Yin, Xiangyu Zhao, Changyao Tian, Yu Qiao, Wenhai Wang, Jifeng Dai, Gen Luo (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-14232.txt.

Description

Zhaokai Wang and seven co-authors build GenExam in 2025, an exam that asks image models to draw rather than just answer. A good exam image needs correct facts, sound reasoning and a clean picture at once. Earlier tests covered understanding or loose drawing, but not strict exam style drawing.

GenExam holds a thousand tasks across ten subjects, sorted under a four level scheme. Each task comes with a correct reference image and a detailed scoring list, so judges can check both meaning and looks. The team tested seventeen image and combined models. All found the exam hard, and open models trailed the leading closed models by a wide margin. The results show how far current systems are from joining knowledge, reasoning and drawing in one step.

Catalog record

CREATORZhaokai Wang, Penghao Yin, Xiangyu Zhao, Changyao Tian, Yu Qiao, Wenhai Wang, Jifeng Dai, Gen Luo
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE74 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2509-14232

Browse the full library · About the Tower of Babel