Multiple-choice question answering on HiCervix n = 8,051 (test)
71.5AccuracyPathChat+
Evaluation results
| Method | Links | |
|---|---|---|
| PathChat+2025.06 | 71.5 | |
| GPT-52025.06 | 45.6 | |
| Gemini 2.5 pro2025.06 | 44.3 | |
| GPT-5-mini2025.06 | 42.6 | |
| LLaVA-OneVision2025.06 | 37.6 | |
| Quilt-LLaVA2025.06 | 37.2 | |
| HuatuoGPT-Vision2025.06 | 37.1 | |
| PathChat 12025.06 | 36.8 | |
| PA-LLaVA2025.06 | 33 | |
| Llama-3.22025.06 | 29.3 | |
| Qwen3-VL2025.06 | 25.8 | |
| Claude Sonnet 42025.06 | 24.9 |