Domains: Science
Skills: AgenticReasoningTool UseCoding

Four-Factor Integrity Breakdown & Primary-Source Evidence

Privacy

Weight 40%
2/10
Access Model: Auto-Gated / Encrypted
"BioMysteryBench is provided for evaluation and benchmarking only. By requesting access you agree that you will not use this benchmark (problem statements, answer rubrics, task formulation) to train, fine-tune, reinforce, or distill any machine-learning model."
Primary source (HuggingFace) ↗ Verified 2026-11-22
"BioMysteryBench was developed by Anthropic. Its task materials are released under CC BY 4.0 with restrictions against their use for model training, while the biological data remain subject to their original repositories’ policies."

Originality

Weight 25%
7.2/10
Reuse Profile: Low Reuse (<25%)
"Tasks are constructed by stripping real-world biological data of metadata (e.g. sample origin) and evaluating whether models can recover this information via computational analysis."

Freshness

Weight 20%
7/10
Initial Release: 2026-04Latest Refresh: 2026-07Rolling Updates: No
"Removed after a June 2026 audit that reran every problem with expert bioinformaticians and multiple model runs, and cross-checked answer keys against the underlying data."
Primary source (HuggingFace) ↗ Verified 2026-11-22

Grading

Weight 15%
3.3/10
Integrity: Self-Reported / Live NetworkSandbox: live internet
"Network access is restricted to the scientific package repositories and biological databases required by the benchmark. Tasks were limited to 200 agent turns and four hours of wall-clock execution."

Frontier Model Standings on BioMysteryBench

Top 15 active models tracked
Rank Model Official Score Percentile
#1Claude Fable 5.2 max Anthropic83.7%100.0p
#2Claude Sonnet 5.5 max Anthropic81.1%100.0p
#3GPT-6.2 Astra max OpenAI80.7%96.6p
#4GPT-6.1 Sol max OpenAI79.6%92.6p
#5Claude Opus 5.5 max Anthropic79.3%86.6p
#5GPT-6 Astra max OpenAI79.3%86.6p
#7GPT-6.2 Sol max OpenAI77.8%80.6p
#8Gemini 4 Argon high Google76.3%76.7p
#9Kimi K3.1 Pro max Kimi75.6%72.6p
#10GPT-6 Sol max OpenAI74.8%68.7p
#11Muse Wonder Full max Meta73.3%64.7p
#12Claude Haiku 5.5 max Anthropic70.4%60.7p
#13Inkling 2 Small default Thinking Machines70.0%56.8p
#14Grok 4.7 xhigh SpaceXAI69.3%48.8p
#14MiMo-V2.6-Flash default Xiaomi69.3%48.8p

Embeddable Audit Badge

View Markdown (.md)

Benchmark maintainers can embed this live SVG badge in GitHub READMEs or documentation:

BenchAudit BioMysteryBench Tier C 4.5/10
[![BenchAudit Tier C](https://benchaudit.org/badges/biomysterybench.svg)](https://benchaudit.org/benchmarks/biomysterybench/)