BioMysteryBench

Biology research-agent benchmark (99-task initial release); per model card footnote evaluated with Claude Code harness, judged by Kimi-K3.

Category

STEM & Reasoning

Max Score

100.0

Score Type

percent

Active

Yes

Model Scores

Scores for BioMysteryBench

Model Score Date Verified
Hy3
54.90
28.08.2026 Verified
Hy4 preview
90.11%
28.08.2026 Verified
DeepSeek V4 Pro
36.81%
28.08.2026 Verified
GLM-5.3
77.47%
28.08.2026 Verified
Kimi K3
35.16%
28.08.2026 Verified
GPT-5.6 Sol
100.00%
28.08.2026 Verified
Claude Opus 5
94.51%
28.08.2026 Verified
Qwen3.8-2.4T-A95B
21.98%
28.08.2026 Verified