Rubric Satisfaction Evaluation on Medical benchmark leaderboard · SOTA2 Research

Rubric Satisfaction Evaluation on Medical

50.9Claude-4 Sonnet ScoreGPT-5-Thinking

Evaluation Results

Method	Links
GPT-5-Thinking2025.12		50.9	69.7	56.5
GPT-OSS-120B2025.12		43.9	63.2	39.1
Claude-4.1-Opus2025.12		43.9	62.2	38.4
Med-Ft-Qwen-3-30B2025.12		43.5	54.4	37
Claude-4-Sonnet2025.12		41.5	58.8	35.2
ML-Ft-Qwen-3-30B2025.12		41.3	50.6	31.6
GPT-OSS-20B2025.12		41	56.9	33
Arxiv-Ft-Qwen-3-30B2025.12		40	51	31.7
Qwen-3-30B-A3B-Thinking2025.12		39.4	49.2	27.6
Qwen-3-30B-A3B2025.12		38.8	50.6	29.3
Grok-4-Fast2025.12		38.8	55.9	34.8
Grok-42025.12		38.5	53.3	34
Gemma-3-27B2025.12		32.5	48.2	24.4
Gemini-2.5-Pro2025.12		32.3	53.3	31.6
Qwen-3-4B2025.12		31	42.4	19.9
Qwen-3-4B-Thinking2025.12		30.7	39.9	17.4
Llama-4-Maverick2025.12		19.8	35.7	16.4
Gemma-3-4B2025.12		18.9	33.9	12.8
Llama-3.3-70B2025.12		18.3	32.7	14.4
Llama-4-Scout2025.12		12.1	20.5	9.2
Llama-3.1-8B2025.12		11	23.4	8.5