AI Assistance Prevention on DOPE Exam Dataset
1Success Rategpt-5.1
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| gpt-5.1Best Method=ICW + Font (gib.)2026.01 | 1 | — | |
| gpt-4oBest Method=ICW + Font (ref.)2026.01 | 0.981 | — | |
| DOPE (Average best hybrid)Description=Average of best hybrid configurations across all tested models2026.01 | 0.963 | — | |
| sonnet 4.5Best Method=ICW + Font (ref.)2026.01 | 0.936 | — | |
| opus 4.5Best Method=ICW + Font (ref.)2026.01 | 0.936 | — | |
| ICW baseline (Average)Description=Average prevention rate for ICW baseline across all tested models2026.01 | 0.632 | — |