Visual Question Answering on SlideBench-VQA BCNB
57.68OverallPathReasoner-R1-7B
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| PathReasoner-R1-7BInput Type=Slide2026.01 | 57.68 | 90.2 | 52.23 | 32.42 | |
| PathMemInput=WSI, Zero-shot=true2026.03 | 57.1 | 91.8 | 48.2 | 31.3 | |
| TC-SSAFlops=1.72T, Token Compression=~ 60×, Zero-shot=true2026.03 | 55.94 | — | — | — | |
| WSI-LLaVA-7BInput Type=Slide2026.01 | 55.3 | 90.5 | 46.3 | 29.2 | |
| WSI-LLaVAInput=WSI, Zero-shot=true2026.03 | 55.3 | 90.5 | 46.3 | 29.2 | |
| SlideChatInput=Slide2024.10 | 54.14 | — | — | — | |
| SlideChatFlops=133.3T, Status=Upper Bound2026.03 | 54.14 | — | — | — | |
| PathReasoner-SFT-7BInput Type=Slide2026.01 | 53.91 | 88.17 | 45.07 | 23.82 | |
| HuatuoGPT-Vision-7BInput Type=Thumbnail2026.01 | 47.5 | 75.43 | 44.82 | 21.93 | |
| SlideChat-7B (Baseline)Input Type=Slide2026.01 | 43.6 | 90.2 | 23.1 | 17.5 | |
| SlideChatInput=WSI, Zero-shot=true2026.03 | 43.6 | 90.2 | 23.1 | 17.5 | |
| Quilt-LLaVAInput=Slide (T)2024.10 | 41.55 | — | — | — | |
| GPT-4oInput=Patch2024.10 | 41.43 | — | — | — | |
| GPT-4oToken Compression=~ 60×, Zero-shot=true2026.03 | 41.43 | — | — | — | |
| InternVL3.5-8BInput Type=Thumbnail2026.01 | 39.02 | 48.87 | 44.38 | 24.48 | |
| MedVLThinker-7BInput Type=Thumbnail2026.01 | 36.42 | 52.08 | 37.9 | 19.47 | |
| MedDrInput=Slide (T)2024.10 | 35.48 | — | — | — | |
| MedGemma-4B-ITInput Type=Thumbnail2026.01 | 35.08 | 56.33 | 28.94 | 19.19 | |
| Qwen2.5-VL-8B-InstructInput Type=Thumbnail2026.01 | 34.06 | 58.13 | 27.11 | 16.07 | |
| MedDrInput=Patch2024.10 | 33.67 | — | — | — | |
| MedDrFlops=1.70T, Token Compression=~ 60×, Zero-shot=true2026.03 | 33.67 | — | — | — | |
| Qwen3-VL-8B-InstructInput Type=Thumbnail2026.01 | 33.43 | 55.01 | 27.65 | 16.92 | |
| Quilt-LLaVAInput=Patch2024.10 | 32.19 | — | — | — | |
| Quilt-LLaVAFlops=1.70T, Token Compression=~ 60×, Zero-shot=true2026.03 | 32.19 | — | — | — | |
| Patho-R1-7BInput Type=Thumbnail2026.01 | 31.43 | 27.13 | 39.42 | 27.13 | |
| Qwen3-VL-8B-ThinkingInput Type=Thumbnail2026.01 | 31.16 | 53.69 | 26.35 | 14.74 | |
| Quilt-LLaVA-7BInput Type=Thumbnail2026.01 | 30.97 | 47.35 | 18.57 | 25.43 | |
| LLaVA-MedInput=Patch2024.10 | 30.1 | — | — | — | |
| LLaVA-MedFlops=1.70T, Token Compression=~ 60×, Zero-shot=true2026.03 | 30.1 | — | — | — | |
| LLaVA-Med-7BInput Type=Thumbnail2026.01 | 26.33 | 31.1 | 39.2 | 10.3 | |
| RandomInput=Text2024.10 | 24.4 | — | — | — | |
| Random BaselineToken Compression=~ 60×, Zero-shot=true2026.03 | 24.4 | — | — | — | |
| Quilt-LLaVAInput=Thumb, Zero-shot=true2026.03 | 16.3 | 11.3 | 19.4 | 18.1 | |
| WSI-VQAInput=WSI, Zero-shot=true2026.03 | 11.3 | 3.9 | 30 | 0 | |
| GPT-4Input=Text2024.10 | 0 | — | — | — | |
| GPT-4oInput=Slide (T)2024.10 | 0 | — | — | — | |
| LLaVA-MedInput=Slide (T)2024.10 | 0 | — | — | — |