Which AI model scores highest on BioMysteryBench?

Gemini 3.8 Flash by Google holds the best BioMysteryBench (hard) result among tracked models, at 56.5% (released Sep 2 2026). Higher scores are better on this benchmark.

What are the top 3 models on BioMysteryBench?

1. Gemini 3.8 Flash (Google) — 56.5%; 2. Claude Opus 5 (Anthropic) — 49.4%; 3. Gemini 3.7 Flash (Google) — 43.5%.

How many models have a published BioMysteryBench score?

3 tracked models have a published BioMysteryBench (hard) score. Scores are the figures reported by each lab at that model's release, so this page is a record of results over time rather than a re-run leaderboard.

Biology

BioMysteryBenchhard

Real unsolved-style biology puzzles — the AI has to reason its way to an answer the way a research biologist would. The “hard” split contains the toughest cases. Higher is better.

hard human solved

Rankings

Higher is better

BioMysteryBench — frequently asked questions

: Real unsolved-style biology puzzles — the AI has to reason its way to an answer the way a research biologist would. The “hard” split contains the toughest cases. Higher is better.
: Gemini 3.8 Flash by Google holds the best BioMysteryBench (hard) result among tracked models, at 56.5% (released Sep 2 2026). Higher scores are better on this benchmark.
: 1. Gemini 3.8 Flash (Google) — 56.5%; 2. Claude Opus 5 (Anthropic) — 49.4%; 3. Gemini 3.7 Flash (Google) — 43.5%.
: 3 tracked models have a published BioMysteryBench (hard) score. Scores are the figures reported by each lab at that model's release, so this page is a record of results over time rather than a re-run leaderboard.