Multiple-choice Visual Question Answering on PathMMU SocialPath n=1,796 (test)
71.7AccuracyPathChat+
Evaluation results
| Method | Links | |
|---|---|---|
| PathChat+2025.06 | 71.7 | |
| GPT-52025.06 | 69.4 | |
| Gemini 2.5 pro2025.06 | 66.3 | |
| GPT-5-mini2025.06 | 65.5 | |
| Claude Sonnet 42025.06 | 62.5 | |
| PathChat 12025.06 | 61.4 | |
| HuatuoGPT-Vision2025.06 | 59.6 | |
| Qwen3-VL2025.06 | 56.1 | |
| Llama-3.22025.06 | 52.1 | |
| LLaVA-OneVision2025.06 | 48.9 | |
| Quilt-LLaVA2025.06 | 45.4 | |
| PA-LLaVA2025.06 | 41 |