Scientific Question Answering on GPQA (AUC Metrics)
4SC Baseline AUCSelf-Consistency
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Self-ConsistencyModel=Gemini 3.1 Pro2026.05 | 4 | — | — | |
| Self-ConsistencyModel=Claude Sonnet 4.62026.05 | 4 | — | — | |
| GEOModel=Gemini 3.1 Pro2026.05 | — | 0.701 | 0.011 | |
| GEOModel=Claude Sonnet 4.62026.05 | — | 0.727 | 0.02 |