Medical Visual Question Answering on OmniMedVQA
88.5AccuracyGMAI-VL
Evaluation Results
| Method | Links | |
|---|---|---|
| GMAI-VLParams=7B, Medical=yes2025.10 | 88.5 | |
| UniMedVLParams=14B, Medical=yes2025.10 | 85.8 | |
| Ours-7BModel Category=Medical Models2026.04 | 83.36 | |
| Lingshu-7BModel Category=Medical Models2026.04 | 82.9 | |
| Lingshu-7BParams=7B, Medical=yes2025.10 | 82 | |
| LINGSHU-7B + MED-SCOUTParameters=7B, Enhancement=Med-Scout2026.01 | 81.9 | |
| LINGSHU-7BParameters=7B2026.01 | 81.4 | |
| Lingshu-32BParams=32B, Medical=yes2025.10 | 80.4 | |
| InternVL3-8BModel Category=General Models2026.04 | 78.45 | |
| INTERNVL3-8BParameters=8B2026.01 | 78.2 | |
| MMedExpert-R1Training Data Size=1000 samples subset2026.01 | 77.93 | |
| MedVLM-R1Training Data Size=80% of dataset2026.01 | 77.38 | |
| GPT-5Model Type=Proprietary2026.01 | 76.9 | |
| Med-R12026.01 | 76.08 | |
| HUATUOGPT-VISION-7B + MED-SCOUTParameters=7B, Enhancement=Med-Scout2026.01 | 75.4 | |
| GEMINI-3-FLASHModel Type=Proprietary2026.01 | 75.3 | |
| HUATUOGPT-VISION-7BParameters=7B2026.01 | 75 | |
| HealthGPT-L14Params=14B, Medical=yes2025.10 | 74.4 | |
| HuatuoGPT-V-7BModel Category=Medical Models2026.04 | 74.2 | |
| Gemini-2.5-FlashParams=Closed, Medical=no2025.10 | 71.33 | |
| BagelParams=14B, Medical=no2025.10 | 71.13 | |
| GPT-4oParams=Closed, Medical=no2025.10 | 71.06 | |
| MEDGEMMA-4B-ITParameters=4B2026.01 | 70.3 | |
| HealthGPT-M3Params=3.8B, Medical=yes2025.10 | 68.5 | |
| MedGemma-4BParams=4B, Medical=yes2025.10 | 68.4 | |
| QWEN2.5-VL-7B-INSTRUCTParameters=7B2026.01 | 63.8 | |
| Qwen2.5-VL-7BModel Category=General Models2026.04 | 63.42 | |
| BioMediX2-8BModel Category=Medical Models2026.04 | 63.3 | |
| Claude Sonnet 4Params=Closed, Medical=no2025.10 | 63.22 | |
| QWEN2.5-VL-3B-INSTRUCTParameters=3B2026.01 | 61.5 | |
| Rform + Rexact + RDTWReward functions=Rform + Rexact + RDTW2026.04 | 55.7 | |
| InternVL2Params=8B, Medical=no2025.10 | 54.5 | |
| Rform + RexactReward functions=Rform + Rexact2026.04 | 52.2 | |
| SFTTraining configuration=SFT2026.04 | 51 | |
| LLaVa1.6-7BModel Category=General Models2026.04 | 50.67 | |
| HuatuoGPT-Vision-7BParams=7B, Medical=yes2025.10 | 50 | |
| QWEN3-VL-4B-INSTRUCT + MED-SCOUTParameters=4B, Enhancement=Med-Scout2026.01 | 48.8 | |
| LLAVA-MED-7BParameters=7B2026.01 | 46.8 | |
| QWEN3-VL-8B-INSTRUCT + MED-SCOUTParameters=8B, Enhancement=Med-Scout2026.01 | 46 | |
| JanusParams=1.3B, Medical=no2025.10 | 45.7 | |
| QWEN3-VL-4B-INSTRUCTParameters=4B2026.01 | 45.5 | |
| LLaVA-v1.5Params=7B, Medical=no2025.10 | 44.7 | |
| LLaVa-Med-7BModel Category=Medical Models2026.04 | 44.3 | |
| QWEN3-VL-8B-INSTRUCTParameters=8B2026.01 | 42.9 | |
| LLaVA-MedParams=7B, Medical=yes2025.10 | 41.3 | |
| RformReward functions=Rform2026.04 | 41 | |
| Med-FlamingoParams=8.3B, Medical=yes2025.10 | 34.9 | |
| RadFMParams=14B, Medical=yes2025.10 | 23.5 |