Atlas

Benchmarks

← All benchmarks

BioMysteryBench — Human-solvable (Vals)

Science · 2026-09-22

The 73 human-solvable bioinformatics tasks, with three Terminus 2 trials per task.

Top models (higher is better)

ModelScore
GPT-6.1 Sol89.0
GPT-6 Astra89.0
Claude Opus 587.7
Kimi K3.1 Pro86.3
Gemini 4 Argon84.9
GPT-6 Sol84.0
Muse Wonder Full82.6
Grok 4.682.2
GPT-5.6 Sol80.8
Claude Haiku 5.579.9
Kimi K379.9
DeepSeek V4.1 Flash78.5
Grok 4.778.5
MiMo-V2.6-Flash78.5
Hy4 Preview76.3
Step 5 Preview76.3
Ember-175.8
Mistral Large 4 Preview75.8
Muse Wonder75.8
DeepSeek-V4-Flash-073174.0
Muse Spark 1.274.0
Gemini 3.8 Flash71.7
GPT-6 Luna71.7
GPT-5.6 Luna69.4
Gemini 3.1 Pro Preview67.6
Gemini 3.6 Flash67.1
Beam 1.0 501B A23B48.9
Loading Atlas data…