Multiple-choice question answering on PathMMU EduContent n=255 (test-tiny)
76.9AccuracyPathChat+
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| PathChat+2025.06 | 76.9 | 71.8 | |
| GPT-52025.06 | 75.7 | 70.6 | |
| GPT-5-mini2025.06 | 75.7 | 70.6 | |
| Gemini 2.5 pro2025.06 | 71.8 | 65.9 | |
| Expert performance2025.06 | 69 | — | |
| PathChat 12025.06 | 67.8 | 61.6 | |
| Claude Sonnet 42025.06 | 66.7 | 61.2 | |
| HuatuoGPT-Vision2025.06 | 62 | 55.7 | |
| Qwen3-VL2025.06 | 61.2 | 55.3 | |
| LLaVA-OneVision2025.06 | 57.3 | 51.4 | |
| Llama-3.22025.06 | 56.5 | 50.6 | |
| Quilt-LLaVA2025.06 | 49.8 | 43.1 | |
| PA-LLaVA2025.06 | 42.4 | 36.1 |