BioMysteryBench contains 99 expert-authored bioinformatics research questions over messy real-world datasets. Agents may use a Linux terminal, bioinformatics software, Python and R, plus restricted access to canonical data sources; answers are graded on objective biological conclusions rather than a prescribed analysis path.
查看 BioMysteryBench 的最新得分、模型模式、发布时间与参数规模,快速了解当前完整榜单表现。
数据优先来自官方发布(GitHub、Hugging Face、论文),其次为评测基准官方结果,最后为第三方评测机构数据。 了解数据收集方法
| 排名 | 模型 | 开源情况 | |||
|---|---|---|---|---|---|
![]() Claude Opus 5.5 思考水平·Max工具 | 79.26 | 2026-09-22 | 未知 | 闭源 | |
![]() GPT-6 Sol 思考水平·Max工具 | 74.81 | 2026-09-22 | 未知 | 闭源 | |
![]() Hy4 preview 思考水平·高工具 | 71.30 | 2026-08-28 | 7700亿 | 免费商用 | |
4 | ![]() GPT-6 Luna 思考水平·Max工具 | 61.48 | 2026-09-22 | 未知 | 闭源 |
Google reported Gemini 3.7 Flash at 87.1% on the 76 human-solvable tasks and 43.5% on the 23 human-difficult tasks in August 2026.