Reasoning on DermoBench Reasoning benchmark leaderboard · SOTA2 Research

Reasoning on DermoBench Reasoning

82.15T3.1 (CoT)Human Performance

Evaluation Results

Method	Links
Human Performance2026.01		82.15	78.41	80.28
DermoGPT-RL + CCT2026.01		67.74	66.64	67.19
DermoGPT-RL2026.01		66.04	65.48	65.76
DermoGPT-SFT + CCT2026.01		63.73	65.31	64.52
DermoGPT-SFT2026.01		62.57	63.34	62.95
Qwen3-VL-32B2026.01		55.04	53.85	54.45
QVQ-72B-Preview2026.01		51.56	54.14	52.85
Gemini-2.5-Flash2026.01		48.92	58.49	53.7
Qwen3-VL-8B (Base)2026.01		47.53	53.43	50.48
Lingshu-7B2026.01		47.16	49.3	48.23
Llama-3.2-90B2026.01		44.61	56.14	50.38
Lingshu-32B2026.01		44.41	49.55	46.98
GLM-4.5V2026.01		44.19	53.28	48.73
Claude-Sonnet-4.5-Thinking2026.01		43.54	54.37	48.95
SkinVL-PubMM2026.01		42.92	54.62	48.77
GPT-4o-mini2026.01		42.83	51.65	47.24
Qwen2.5-VL-72B2026.01		40.39	49.71	45.05
HuatuoGPT-Vis-7B2026.01		39.41	43.98	41.69
LLaVA-Med-v1.52026.01		38.33	46.19	42.26
Llama-3.2-11B2026.01		36.16	38.29	37.22
Nemotron-Nano2026.01		31.9	37.4	34.65