This BioMysteryBench subset contains the 23 tasks that no member of the human-expert panel solved correctly. Tasks still have controlled objective answers and use real-world bioinformatics data, tools and restricted external data access.
Browse the latest scores, model modes, release dates, and parameter sizes for BioMysteryBench (Human-difficult).
Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology
| Rank | Model | License | |||
|---|---|---|---|---|---|
 Gemini 3.7 Flash Thinking Level · MediumToolsInternet | 43.50 | 2026-08-13 | Unknown | Closed |
Google reported 43.5% for Gemini 3.7 Flash on the BioMysteryBench human-difficult subset in August 2026.
