QC-Bench

QC-Bench is a comprehensive quantum computing benchmark for evaluating LLMs, containing 6,021 questions (5,184 multiple-choice, 416 true/false, and 421 open-ended) covering seven core quantum computing areas. The benchmark includes two human-authored subsets (QC200 and QC1000) derived from four decades of research literature, plus 4,184 AI-extracted questions, and has been used to evaluate 31 proprietary and open-source LLMs.

QC-Bench 是一款用于评估大语言模型（Large Language Model）的综合性量子计算基准测试集，共收录6021道问题（其中5184道选择题、416道判断题及421道开放式问答题），覆盖七大核心量子计算领域。该基准测试集包含两个人工编写子集QC200与QC1000，其数据源自四十年间的研究文献，此外还包含4184道由人工智能提取的问题，目前已被用于评估31款专有与开源大语言模型。