List your company

Scientific Question Answering on GPQA (AUC Metrics)

4SC Baseline AUCSelf-Consistency
3.83.944.1May 7, 2026

Evaluation results

MethodLinks
Self-ConsistencyModel=Gemini 3.1 Pro2026.054——
Self-ConsistencyModel=Claude Sonnet 4.62026.054——
GEOModel=Gemini 3.1 Pro2026.05—0.7010.011
GEOModel=Claude Sonnet 4.62026.05—0.7270.02