TableBench

About TableBench

TableBench is a comprehensive and complex benchmark designed to evaluate Table Question Answering (TableQA) capabilities, aligning closely with the "Reasoning Complexity of Questions" dimension in real-world Table QA scenarios. It covers 18 question categories across 4 major ategories—including Fact Checking, Numerical Reasoning, Data Analysis, and Visualization—with 886 carefully curated test cases. TableBench substantially pushes the boundaries of large language models in complex TableQA scenarios.

News

Challenges from TableBench

1. Multi-hop fact-checking involves multiple steps to establish the relationship between facts across information from table.

2. Multi-hop numerical reasoning involves calculating the intermediate results of multiple steps based on the data in the table to achieve the final conclusion.

3. Trend forecasting involves estimating future data trends based on historical data analysis.

4. Chart generation necessitates executing program commands to create charts.

Submission

🤗🤗 We warmly welcome submissions to our leaderboard, including both your own methods and contributions showcasing the latest model performance! TableBench features two separate leaderboards. Please refer to the Submission Guidelines below for details, and submit your results as instructed to tablebench2025@gmail.com.

Citation

@inproceedings{wu2025tablebench,
  title={Tablebench: A comprehensive and complex benchmark for table question answering},
  author={Wu, Xianjie and Yang, Jian and Chai, Linzheng and Zhang, Ge and Liu, Jiaheng and Du, Xeron and Liang, Di and Shu, Daixin and Cheng, Xianfu and Sun, Tianzhen and others},
  booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
  volume={39},
  number={24},
  pages={25497--25506},
  year={2025}
}

⚙ This website is an improved version based on the original source code from Bird-Bench!