Multiple-choice Visual Question Answering on PathMMU SocialPath tiny n=229 (test)
75.1AccuracyPathChat+
Evaluation Results
| Method | Links | |
|---|---|---|
| PathChat+2025.06 | 75.1 | |
| GPT-52025.06 | 73.8 | |
| Expert performance2025.06 | 71.5 | |
| Gemini 2.5 pro2025.06 | 70.3 | |
| Claude Sonnet 42025.06 | 66.8 | |
| GPT-5-mini2025.06 | 66.8 | |
| Qwen3-VL2025.06 | 64.2 | |
| PathChat 12025.06 | 63.8 | |
| HuatuoGPT-Vision2025.06 | 57.6 | |
| Llama-3.22025.06 | 52 | |
| LLaVA-OneVision2025.06 | 48.9 | |
| PA-LLaVA2025.06 | 45.4 | |
| Quilt-LLaVA2025.06 | 44.1 |