Relevance Classification on AR-Med benchmark leaderboard · SOTA2 Research

Relevance Classification on AR-Med

98.54Highly Relevant PrecisionGPT-4o

Evaluation Results

Method	Links
GPT-4o2025.12		98.54	91.1	94.66	21.49	61.29	32.09	71.43	32.44	44.62
Qwen3-32B2025.12		98.39	94.29	96.3	30.61	61.59	41.09	63.87	32.81	43.29
QwQ-32B2025.12		97.84	95.3	96.56	36.73	54.97	43.98	65.55	27.46	38.7
Qwen2-72B2025.12		97.15	95.49	96.31	32.18	46.77	38.18	55.46	39.76	46.38
Llama3-70B2025.12		96.95	96.64	96.79	52.81	45.52	48.88	26.05	46.27	33.39
Deepseek-r1-70B2025.12		96.89	96.33	96.61	37.88	43.1	40.32	64.71	37.38	47.41