Diagnostic Classification on derm7pt Melanoma vs. Atypical Nevus (test)
77.6AccuracyCARE
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| CAREBase Model=Gemini-3-Flash, Strategy=CARE2026.02 | 77.6 | 76.9 | 55.2 | |
| Gemini-3-Pro2026.02 | 74.1 | 73.4 | 48.2 | |
| Blind-CAREBase Model=Gemini-3-Flash, Strategy=Blind-CARE, Access to original image=false2026.02 | 73.9 | 72.9 | 47.6 | |
| Gemini-3-Flash (Self-Check 3x)Base Model=Gemini-3-Flash, Strategy=Self-Check, Number of passes=3x, Multi-pass self-revision=true2026.02 | 71.9 | 70.4 | 43.6 | |
| GLM-4.6V-Flash2026.02 | 71.8 | 71.4 | 42.6 | |
| Gemini-3-Flash (Self-Check 2x)Base Model=Gemini-3-Flash, Strategy=Self-Check, Number of passes=2x, Multi-pass self-revision=true2026.02 | 70.1 | 68.1 | 40 | |
| Qwen3-VL-32BParameters=32B2026.02 | 69.8 | 67 | 34.7 | |
| Gemini-3-Flash (Majority-Vote 3x)Base Model=Gemini-3-Flash, Strategy=Majority-Vote, Number of sampling=3x, Independent sampling with majority aggregation=true2026.02 | 68.6 | 66 | 37.6 | |
| Gemini-3-Flash (Baseline)Base Model=Gemini-3-Flash, Strategy=Baseline2026.02 | 66.5 | 63 | 32.8 | |
| Qwen3-VL-8BParameters=8B2026.02 | 66.1 | 65.6 | 36.5 | |
| SigLIP22026.02 | 61.1 | 60.6 | 21.9 | |
| BiomedCLIP2026.02 | 60.7 | 56.9 | 20.9 | |
| InternVL3-14BParameters=14B2026.02 | 54.9 | 53.4 | 7.1 | |
| MedVLM-R12026.02 | 50.5 | 36.8 | 0.1 | |
| Gemma-3-12BParameters=12B2026.02 | 49.9 | 42.7 | 9.1 |