Long-form QA on MultiMed-X SW
4.55Overall ScoreMED-COREASONER
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| MED-COREASONERbackbone=GPT-5.12026.01 | 4.55 | 4.57 | 4.8 | 4.82 | 4.47 | 94.5 | |
| GPT-5.22026.01 | 4.34 | 4.415 | 4.59 | 4.795 | 4.345 | 89 | |
| GPT-5.12026.01 | 4.33 | 4.385 | 4.545 | 4.805 | 4.255 | 88.5 | |
| GPT-4o2026.01 | 4.04 | 4.07 | 4.115 | 4.755 | 4.155 | 81.5 | |
| Claude-3.5-haiku2026.01 | 3.825 | 3.855 | 3.95 | 4.565 | 3.78 | 73 |