titleResearchClawBench: A Benchmark for End-to-End Autonomous Scientific Research
authorsWanghan Xu, Shuo Li, Tianlin Ye, Qinglong Cao, Yixin Chen, Hengjian Gao, Yiheng Wang, Qi Li, Kun Li, Sheng Xu, Shengdu Chai, Fangchen Yu, Xiangyu Zhao, Zhangrui Zhao, Weijie Ma, Zijie Guo, Haoyu Zhou, Haoxiang Yin, Lixue Cheng, Chaofan Hu, Haoxuan Li, Lu Mi, Xuxuan Xie, Yifan Zhou, Ruizhe Chen, Zhiwang Zhou, Xingjian Guo, Yuhao Zhou, Xuming He, Shengyuan Xu, Xinyu Gu, Jiamin Wu, Mianxin Liu, Chunfeng Song, Fenghua Ling, Dongzhan Zhou, Shixiang Tang, Yuqiang Li, Mao Su, Peng Ye, Siqi Sun, Bin Wang, Xue Yang, Zhenfei Yin, Tianfan Fu, Guangtao Zhai, Wanli Ouyang, Bo Zhang, Lei Bai, Wenlong Zhang
date2026-05-28
venuearXiv (preprint, not peer reviewed)
idarXiv:2606.07591
affiliationIntern Science
upvotes106 (Hugging Face Daily Papers, 2026-11-01)
linksabs pdf hf code
abstractAI coding agents are increasingly used for scientific work, but their end-to-end autonomous research capability remains difficult to verify. We present ResearchClawBench, a benchmark for evaluating autonomous scientific research across 40 tasks from 10 scientific domains. Each task is grounded in a real published paper, provides related literature and raw data, and hides the target paper during evaluation. Expert-curated multimodal rubrics decompose the target scientific artifacts into weighted criteria, enabling evaluation of target-paper-level re-discovery while leaving room for new discovery. We evaluate seven autonomous research (auto-research) agents under a unified protocol and seventeen native LLMs through the lightweight ResearchHarness. Current systems remain far from reliable re-discovery: the strongest autonomous agent, Claude Code, averages 21.5, and the strongest ResearchHarness LLM, Claude-Opus-4.7, averages 20.7, with an LLM frontier mean of only 26.5. Error analysis shows that failures concentrate in experimental protocol mismatch, evidence mismatch, and missing scientific core. ResearchClawBench provides a reproducible evaluation frontier for measuring progress toward autonomous scientific research.

cite

@misc{xu2026researchclawbench,
  title={ResearchClawBench: A Benchmark for End-to-End Autonomous Scientific Research},
  author={Wanghan Xu and Shuo Li and Tianlin Ye and Qinglong Cao and Yixin Chen and Hengjian Gao and Yiheng Wang and Qi Li and Kun Li and Sheng Xu and Shengdu Chai and Fangchen Yu and Xiangyu Zhao and Zhangrui Zhao and Weijie Ma and Zijie Guo and Haoyu Zhou and Haoxiang Yin and Lixue Cheng and Chaofan Hu and Haoxuan Li and Lu Mi and Xuxuan Xie and Yifan Zhou and Ruizhe Chen and Zhiwang Zhou and Xingjian Guo and Yuhao Zhou and Xuming He and Shengyuan Xu and Xinyu Gu and Jiamin Wu and Mianxin Liu and Chunfeng Song and Fenghua Ling and Dongzhan Zhou and Shixiang Tang and Yuqiang Li and Mao Su and Peng Ye and Siqi Sun and Bin Wang and Xue Yang and Zhenfei Yin and Tianfan Fu and Guangtao Zhai and Wanli Ouyang and Bo Zhang and Lei Bai and Wenlong Zhang},
  year={2026},
  eprint={2606.07591},
  archivePrefix={arXiv},
  url={https://arxiv.org/abs/2606.07591}
}