Medical Visual Question Answering on PathVQA
85.37AccuracyViToS
Evaluation Results
| Method | Links | |
|---|---|---|
| ViToSBackbone=Lingshu-32B2026.06 | 85.37 | |
| VisionZipBackbone=Lingshu-32B2026.06 | 77.75 | |
| VScanBackbone=Lingshu-32B2026.06 | 77.36 | |
| PDropBackbone=Lingshu-32B2026.06 | 77.31 | |
| BaselineBackbone=Lingshu-32B2026.06 | 77.28 | |
| ViToS2026.06 | 77.2 | |
| ViToSBackbone=Lingshu-7B2026.06 | 77.19 | |
| FastVBackbone=Lingshu-32B2026.06 | 77.16 | |
| TMA-AllComponBase Model=GPT-4o2025.08 | 76.8 | |
| VisionZipBackbone=Lingshu-7B2026.06 | 73.26 | |
| VScanBackbone=Lingshu-7B2026.06 | 73.26 | |
| PDropBackbone=Lingshu-7B2026.06 | 73.17 | |
| BaselineBackbone=Lingshu-7B2026.06 | 71.98 | |
| FastVBackbone=Lingshu-7B2026.06 | 71.48 | |
| Rform + Rexact + RDTWReward functions=Rform + Rexact + RDTW2026.04 | 71 | |
| DyLANBase Model=GPT-4o2025.08 | 69.7 | |
| TMA-AllComponBase Model=MedGemma-4B2025.08 | 68.4 | |
| ViTAR2026.06 | 67 | |
| MedVLThinker-7BModel Type=Medical LLM2025.10 | 66.83 | |
| MedAgentsBase Model=GPT-4o2025.08 | 66.7 | |
| Med-R1-2BParameters=2B2026.03 | 66.5 | |
| Rform + RexactReward functions=Rform + Rexact2026.04 | 66.4 | |
| VisionZipBackbone=HuatuoGPT-Vision-7B2026.06 | 66 | |
| MedVLThinker2026.06 | 66 | |
| BaselineBackbone=HuatuoGPT-Vision-7B2026.06 | 65.76 | |
| VScanBackbone=HuatuoGPT-Vision-7B2026.06 | 65.68 | |
| MedVLSynther-7BModel Type=Medical LLM2025.10 | 65.56 | |
| Ours (7B)Parameters=7B2026.03 | 65.5 | |
| ReConcileBase Model=GPT-4o2025.08 | 65.4 | |
| Qwen2.5-VL-7B-InstructModel Type=General LLM2025.10 | 65.39 | |
| MDAgentsBase Model=GPT-4o2025.08 | 65.3 | |
| Qwen2.5-VL-7B-InstructParameters=7B2026.03 | 65 | |
| FastVBackbone=HuatuoGPT-Vision-7B2026.06 | 64.93 | |
| RformReward functions=Rform2026.04 | 64.3 | |
| PDropBackbone=HuatuoGPT-Vision-7B2026.06 | 64.04 | |
| ReConcileBase Model=Gemma-3-4B2025.08 | 63.8 | |
| ViToSBackbone=HuatuoGPT-Vision-7B2026.06 | 63.8 | |
| HuatuoGPT-Vision-7BModel Type=Medical LLM2025.10 | 63.53 | |
| QoQ-Med-7BParameters=7B2026.03 | 63.5 | |
| OctoMed-7BParameters=7B2026.03 | 63 | |
| MedVLSynther-3BModel Type=Medical LLM2025.10 | 62.82 | |
| MedVLThinker-3BModel Type=Medical LLM2025.10 | 62.28 | |
| MedGemma 27BModel Type=Medical LLM2025.10 | 62.09 | |
| Citrus-VSize=8B2026.03 | 62 | |
| Qwen2.5-VL-3B-InstructModel Type=General LLM2025.10 | 61.96 | |
| LingshuSize=7B2026.03 | 61.9 | |
| Lingshu-7BModel Category=Medical Models2026.04 | 61.9 | |
| MedVLM-R1-2BParameters=2B2026.03 | 61.5 | |
| Lingshu-32BParams=32B, Medical=yes2025.10 | 61.3 | |
| MedAgentsBase Model=Gemma-3-4B2025.08 | 60.2 | |
| TMA-AllComponBase Model=Gemma-3-4B2025.08 | 59.7 | |
| Gemini-2.5-FlashParams=Closed, Medical=no2025.10 | 59.7 | |
| MedGemma 4BModel Type=Medical LLM2025.10 | 59.64 | |
| Lingshu-7BParams=7B, Medical=yes2025.10 | 59.6 | |
| Gemme 3 4BModel Type=General LLM2025.10 | 59.24 | |
| MedVLThinker-7BParameters=7B2026.03 | 57.5 | |
| Llava Med V1.5 7BModel Type=Medical LLM2025.10 | 56.52 | |
| MEDIC-ADSize=7B2026.03 | 56.5 | |
| MedVLM-R12026.06 | 55.2 | |
| GPT-4oParams=Closed, Medical=no2025.10 | 54.5 | |
| Claude Sonnet 4Params=Closed, Medical=no2025.10 | 54.2 | |
| UniMedVLParams=14B, Medical=yes2025.10 | 53.5 | |
| Med-R12026.06 | 53.3 | |
| SFTTraining configuration=SFT2026.04 | 51.4 | |
| ToolTreeBackbone Model=GPT-4o2026.03 | 50.86 | |
| MDAgentsBase Model=Gemma-3-4B2025.08 | 49.1 | |
| HuatuoGPT-V-7BModel Category=Medical Models2026.04 | 48 | |
| DyLANBase Model=Gemma-3-4B2025.08 | 47.4 | |
| ToolTreeBackbone Model=GPT-4o-mini2026.03 | 47.12 | |
| Ours-7BModel Category=Medical Models2026.04 | 46.5 | |
| OctoToolsBackbone Model=GPT-4o2026.03 | 46.17 | |
| LingshuSize=7B, Training protocol=full finetuning2026.03 | 45.3 | |
| HealthGPT-L14Params=14B, Medical=yes2025.10 | 44.4 | |
| OctoToolsBackbone Model=GPT-4o-mini2026.03 | 43.13 | |
| MedRouteModel=Ours2026.02 | 41.3 | |
| InternVL3-8BModel Category=General Models2026.04 | 40.79 | |
| HuggingGPTBackbone Model=GPT-4o-mini2026.03 | 40.72 | |
| GPT-4.1-miniModel=GPT-4.1-mini2026.02 | 40.59 | |
| LLaVA-Med-v1.5-Mistral-7BModel=LLaVA-Med-v1.5-Mistral-7B2026.02 | 40.56 | |
| GMAI-VLParams=7B, Medical=yes2025.10 | 39.8 | |
| HealthGPT-M3Params=3.8B, Medical=yes2025.10 | 39.7 | |
| BagelParams=14B, Medical=no2025.10 | 39.05 | |
| Qwen2.5-VL-7BModel Category=General Models2026.04 | 38.8 | |
| LLaVa-Med-7BModel Category=Medical Models2026.04 | 38.8 | |
| MAMModel=MAM2026.02 | 38.37 | |
| HuggingGPTBackbone Model=GPT-4o2026.03 | 37.82 | |
| BioMediX2-8BModel Category=Medical Models2026.04 | 37 | |
| LLaVA-MedParams=7B, Medical=yes2025.10 | 35.7 | |
| MedGemma-4BParams=4B, Medical=yes2025.10 | 33.7 | |
| LLaVA-MedSize=7B2026.03 | 32.5 | |
| HuatuoGPT-Vision-7BParams=7B, Medical=yes2025.10 | 32 | |
| InternVL2Params=8B, Medical=no2025.10 | 31.9 | |
| LLaVA-v1.5Params=7B, Medical=no2025.10 | 31.4 | |
| Med-FlamingoParams=8.3B, Medical=yes2025.10 | 31.3 | |
| LLaVa1.6-7BModel Category=General Models2026.04 | 29.57 | |
| JanusParams=1.3B, Medical=no2025.10 | 27.9 | |
| Few-ShotBackbone Model=GPT-4o2026.03 | 26.2 | |
| Few-ShotBackbone Model=GPT-4o-mini2026.03 | 24.9 | |
| Qwen2.5-VLModel=Qwen2.5-VL2026.02 | 19.46 | |
| LLaVA-OneVisionModel=LLaVA-OneVision2026.02 | 17.93 |