Atlas

Benchmarks

← All benchmarks

LAB-Bench FigQA

Science · 2024-07-14

FigQA is the LAB-Bench (FutureHouse) task in which models answer multiple-choice questions that require interpreting figures from biology research papers. Scores report accuracy.

Top models (higher is better)

ModelScore
Grok 4.2066.0
Grok 4.134.0
Grok 429.0
Loading Atlas data…