Multimodal Medical Reasoning on MIMIC-CXR-VQA
76.28AccuracyDialectic-Med
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Dialectic-MedBackbone=GPT-5.12026.04 | 76.28 | 4,500 | 22.5 | |
| Dialectic-MedBackbone=Gemini-3 Pro2026.04 | 74.8 | 4,000 | 16 | |
| MedMMVCategory=Medical Agents2026.04 | 73.2 | 9,000 | 36 | |
| Dialectic-MedBackbone=GPT-4o2026.04 | 72.46 | 3,800 | 15.2 | |
| Dialectic-MedBackbone=Claude-4.5 Sonnet2026.04 | 69.85 | 4,200 | 22.68 | |
| GPT-5.1Category=CoT Baseline2026.04 | 68.1 | 1,200 | 6 | |
| Dialectic-MedBackbone=Qwen3-VL-32B2026.04 | 66.81 | 3,000 | 0.35 | |
| Gemini-3 ProCategory=CoT Baseline2026.04 | 66.42 | 1,100 | 4.4 | |
| MedVCDCategory=Medical Agents2026.04 | 66.19 | 8,500 | 34 | |
| Med-PaLM 2Category=CoT Baseline2026.04 | 65.81 | 800 | 1.5 | |
| GPT-4oCategory=CoT Baseline2026.04 | 65.03 | 1,000 | 4 | |
| Claude-4.5 SonnetCategory=CoT Baseline2026.04 | 63.81 | 1,200 | 6.48 | |
| ReConcileCategory=Medical Agents2026.04 | 62.31 | 12,000 | 48 | |
| Dialectic-MedBackbone=Qwen3-VL-8B2026.04 | 61.15 | 2,500 | 0.15 | |
| MedAgentCategory=Medical Agents2026.04 | 58.09 | 15,000 | 60 | |
| Qwen3-VL-32BCategory=CoT Baseline2026.04 | 58.01 | 900 | 0.1 | |
| Qwen3-VL-8BCategory=CoT Baseline2026.04 | 52.21 | 800 | 0.05 | |
| LLaVA-MedCategory=CoT Baseline2026.04 | 48.21 | 700 | 0.05 |