MCQ Diagnostic Accuracy on Stanford Echo (test)
64AccuracyMARCUS
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| MARCUSN=502026.03 | 64 | 50 | — | |
| GPT-5N=502026.03 | 34 | 22 | — | |
| Gemini 2.5 ProN=482026.03 | 22.9 | 10.4 | — | |
| MARCUS vs Gemini (McNemar p)N=482026.03 | — | — | 0.001 | |
| MARCUS vs GPT-5 (McNemar p)N=502026.03 | — | — | 0.007 |