Visual Reasoning on DeepChestVQA
38Acc (Largest Lesion Diameter)GPT-5
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| GPT-52026.02 | 38 | 18 | 30 | 55 | 38 | 38 | 55 | 45 | 38 | |
| 3DMedAgentAgent Model=GPT-52026.02 | 38 | 38 | 40 | 58 | 45 | 63 | 60 | 52 | 49 | |
| 3DMedAgentAgent Model=Qwen3-VL2026.02 | 32 | 40 | 33 | 52 | 42 | 63 | 57 | 43 | 45 | |
| RadFM2026.02 | 28 | 20 | 20 | 35 | 22 | 50 | 43 | 23 | 30 | |
| Qwen3-VLModel Parameters=30B2026.02 | 27 | 38 | 25 | 27 | 13 | 50 | 45 | 32 | 32 | |
| Huatuo-GPTModel Parameters=34B2026.02 | 27 | 27 | 12 | 23 | 18 | 52 | 55 | 15 | 29 | |
| Rand2026.02 | 25 | 20 | 25 | 25 | 25 | 50 | 50 | 25 | 31 | |
| M3D2026.02 | 22 | 17 | 17 | 33 | 30 | 58 | 50 | 27 | 32 | |
| MedGemmaModel Parameters=27B2026.02 | 10 | 23 | 40 | 27 | 20 | 50 | 52 | 23 | 31 |