Medical Reasoning on OPENMEDREASON-Bench
74.3Perception PresenceOPENMEDREASON
Evaluation results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| OPENMEDREASONBackbone=Qwen2.5-VL 7B2026.06 | 74.3 | 87.8 | 65.2 | 89.2 | 95.3 | 85 | 87.9 | 94.5 | 83.1 | 77.8 | 78.5 | |
| gemini-3-flashModel Type=closed-source frontier model2026.06 | 70.8 | 91.2 | 64.6 | 73.8 | 94.3 | 69.6 | 80.4 | 96.1 | 77.3 | 70.5 | 72 | |
| gpt-5-miniModel Type=closed-source frontier model2026.06 | 67.9 | 91.1 | 61.9 | 76.4 | 92.6 | 70.8 | 83.6 | 95.9 | 80.2 | 70.9 | 67.7 | |
| claude-haiku-4.5Model Type=closed-source frontier model2026.06 | 62.8 | 84.2 | 52.9 | 80.7 | 93 | 75.1 | 79.9 | 92 | 73.5 | 67.1 | 57.3 | |
| OctoMed 7BModel Type=open-weight baseline, Parameters=7B2026.06 | 58.4 | 81.2 | 47.4 | 76.1 | 89.8 | 68.3 | 72.7 | 87.4 | 63.6 | 59.8 | 51.2 | |
| Qwen2.5-VL 7BModel Type=open-weight baseline, Parameters=7B2026.06 | 45.1 | 76 | 34.3 | 47.2 | 73 | 34.5 | 58.9 | 73.8 | 43.5 | 37.4 | 47.1 |