Visual Question Answering on PathMMU Tiny 1.0 (test)
72.4Overall AccuracyCPath-Omni
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| CPath-OmniModel Domain=Pathology-specific2024.12 | 72.4 | 74 | 76.6 | 69.8 | 65.9 | 75.7 | |
| Expert performance2024.12 | 71.8 | 72.9 | 71.5 | 69 | 68.3 | 78.9 | |
| Expert PerformanceType=Human Expert2026.01 | 71.8 | 72.9 | 71.5 | 69 | 68.3 | 78.9 | |
| Patho-R1-7BReasoning Ability=Reasoning, Parameters=7B2026.01 | 69.53 | 72.24 | 67.89 | 75.29 | 81.73 | 44.63 | |
| Patho-R1-7BModel Size=Large, Thinking Ability=Yes2025.05 | 69.53 | 72.24 | 67.89 | 75.29 | 81.73 | 44.63 | |
| PathReasoner-R1-7BReasoning Ability=Reasoning, Parameters=7B, Training Strategy=R12026.01 | 67.7 | 69.39 | 68.08 | 70.98 | 73.55 | 56.49 | |
| Patho-R1-3BModel Size=Small, Thinking Ability=Yes2025.05 | 63.83 | 67.62 | 66.51 | 67.45 | 74.52 | 36.72 | |
| PathGen-LLaVA-13BModel Size=Large, Thinking Ability=No2025.05 | 61.9 | 62.28 | 59.63 | 62.35 | 68.27 | 55.93 | |
| PathReasoner-SFT-7BReasoning Ability=Reasoning, Parameters=7B, Training Strategy=SFT2026.01 | 61.19 | 65.83 | 63.4 | 66.66 | 64.9 | 45.19 | |
| PathGen-LLaVAModel Domain=Pathology-specific2024.12 | 60.1 | 60.1 | 60.9 | 60.8 | 63.5 | 54.2 | |
| HuatuoGPT-Vision-7BReasoning Ability=Non-reasoning, Parameters=7B2026.01 | 58.21 | 61.92 | 58.72 | 60 | 65.87 | 40.11 | |
| HuatuoGPT-Vision-7BModel Size=Large, Thinking Ability=No2025.05 | 58.21 | 61.92 | 58.72 | 60 | 65.87 | 40.11 | |
| InternVL3.5-8BReasoning Ability=Reasoning, Parameters=8B2026.01 | 55.4 | 57.65 | 60.55 | 54.9 | 58.17 | 42.94 | |
| InternVL3-8BModel Size=Large, Thinking Ability=Yes2025.05 | 55.4 | 57.65 | 60.55 | 54.9 | 58.17 | 42.94 | |
| GPT-4V-1106Model Domain=General2024.12 | 53.9 | 59.4 | 58.7 | 60.4 | 48.1 | 36.2 | |
| InternVL2.5-8BModel Size=Large, Thinking Ability=Yes2025.05 | 50.13 | 55.87 | 57.8 | 50.59 | 51.44 | 29.38 | |
| Qwen-VL-MAXModel Domain=General2024.12 | 49.2 | 53 | 53.6 | 52.2 | 51.4 | 30.5 | |
| SlideChat-7BReasoning Ability=Non-reasoning, Parameters=7B2026.01 | 46.01 | 53.38 | 45.41 | 47.06 | 49.04 | 29.94 | |
| MedVLThinker-7BReasoning Ability=Reasoning, Parameters=7B2026.01 | 46.01 | 53.38 | 45.41 | 47.06 | 49.04 | 29.94 | |
| Llama-3.2V-11B-cotModel Size=Large, Thinking Ability=Yes2025.05 | 46.01 | 53.38 | 45.41 | 47.06 | 49.04 | 29.94 | |
| Quilt-LLaVAModel Domain=Pathology-specific2024.12 | 45.6 | 47.3 | 46.4 | 51.8 | 46.2 | 32.2 | |
| InternVL2-8BModel Size=Large, Thinking Ability=Yes2025.05 | 45.13 | 49.11 | 51.38 | 50.59 | 46.63 | 21.47 | |
| WSI-LLaVA-7BReasoning Ability=Non-reasoning, Parameters=7B2026.01 | 44.83 | 51 | 46 | 46.67 | 50 | 28.5 | |
| Qwen2.5VL-3BModel Size=Small, Thinking Ability=Yes2025.05 | 43.37 | 49.82 | 46.79 | 44.31 | 46.63 | 23.73 | |
| Gemini Pro VisionModel Domain=General2024.12 | 42.8 | 43.8 | 42.4 | 43.5 | 49.5 | 32.8 | |
| Qwen3-VL-8B-ThinkingReasoning Ability=Reasoning, Parameters=8B, Training Strategy=Thinking2026.01 | 41.88 | 44.84 | 40.83 | 49.41 | 44.23 | 24.86 | |
| Qwen2.5VL-7BModel Size=Large, Thinking Ability=Yes2025.05 | 41.88 | 44.84 | 40.83 | 49.41 | 44.23 | 24.86 | |
| Llama-3.2-11B-Vision-InstructModel Size=Large, Thinking Ability=Yes2025.05 | 39.07 | 39.5 | 41.74 | 38.04 | 45.19 | 29.38 | |
| LLaVA-1.5-13BModel Domain=General2024.12 | 38.8 | 44.5 | 40.4 | 34.1 | 47.1 | 24.9 | |
| InstructBLIP-FLAN-T5-XXLModel Domain=General2024.12 | 34.3 | 39.1 | 33.6 | 34.5 | 38.5 | 22.6 | |
| DeepSeek-VL2-TinyModel Size=Small, Thinking Ability=Yes2025.05 | 34.24 | 38.79 | 36.7 | 33.73 | 38.46 | 19.77 | |
| Quilt-LLaVA-7BReasoning Ability=Non-reasoning, Parameters=7B2026.01 | 33.98 | 37.01 | 32.57 | 38.43 | 42.79 | 14.12 | |
| Quilt-LLaVA-v1.5-7BModel Size=Large, Thinking Ability=No2025.05 | 33.98 | 37.01 | 32.57 | 38.43 | 42.79 | 14.12 | |
| VILA1.5-3BModel Size=Small, Thinking Ability=No2025.05 | 27.57 | 34.88 | 23.85 | 29.41 | 28.85 | 16.38 | |
| MedGemma-4B-ITReasoning Ability=Non-reasoning, Parameters=4B, Training Strategy=IT2026.01 | 26.65 | 28.8 | 31.1 | 29.1 | 33.15 | 8.05 | |
| Llama3.2-3B-vlModel Size=Small, Thinking Ability=No2025.05 | 25.55 | 27.76 | 30.28 | 27.84 | 30.29 | 7.34 | |
| LLaVA-MedModel Domain=Pathology-specific2024.12 | 25.3 | 28.5 | 28.9 | 22.7 | 22.6 | 22.6 | |
| LLaVA-Med-7BReasoning Ability=Non-reasoning, Parameters=7B2026.01 | 23.79 | 31.32 | 18.35 | 21.18 | 31.25 | 13.56 | |
| LLaVA-Onevision-7BModel Size=Large, Thinking Ability=Yes2025.05 | 23.79 | 31.32 | 18.35 | 21.18 | 31.25 | 13.56 | |
| Qwen3-VL-8B-InstructReasoning Ability=Non-reasoning, Parameters=8B, Training Strategy=Instruct2026.01 | 20.9 | 24.56 | 22.48 | 21.57 | 25 | 7.34 | |
| LLaVA-Med-v1.5-Mistral-7BModel Size=Large, Thinking Ability=No2025.05 | 20.9 | 24.56 | 22.48 | 21.57 | 25 | 7.34 | |
| Paligemma2-3b-pt-224Model Size=Small, Thinking Ability=No2025.05 | 1.32 | 2.14 | 0.92 | 1.96 | 0.96 | 0 |