Long-form QA on MultiMed-X TH
4.66Overall ScoreMED-COREASONER
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| MED-COREASONERbackbone=GPT-5.12026.01 | 4.66 | 4.67 | 4.88 | 4.91 | 4.62 | 96 | |
| GPT-5.12026.01 | 4.49 | 4.535 | 4.645 | 4.83 | 4.545 | 91 | |
| GPT-5.22026.01 | 4.44 | 4.49 | 4.605 | 4.835 | 4.43 | 90 | |
| GPT-4o2026.01 | 4.16 | 4.215 | 4.18 | 4.89 | 4.28 | 85.5 | |
| Claude-3.5-haiku2026.01 | 4.135 | 4.205 | 4.2 | 4.79 | 4.33 | 85.5 |