LLM Evaluation on JudgeBench (test) benchmark leaderboard · SOTA2 Research

LLM Evaluation on JudgeBench (test)

79.9KnowledgeSkywork-Reward-V2-Llama-3.1-8B-40M

Evaluation Results

Method	Links
Skywork-Reward-V2-Llama-3.1-8B-40M2025.07		79.9	78.6	89.3	85.7	83.4
Skywork-Reward-V2-Llama-3.1-8B2025.07		76.6	75.5	89.3	78.6	80
Skywork-Reward-V2-Qwen3-8B2025.07		70.1	67.3	82.1	73.8	73.3
o3-mini2025.07		67.5	89.8	87.5	100	86.2
Skywork-Reward-V2-Qwen3-1.7B2025.07		66.9	69.4	83.9	71.4	72.9
Skywork-Reward-V2-Qwen3-4B2025.07		66.9	64.3	80.4	66.7	69.5
o1-preview2025.07		66.2	79.6	85.7	85.7	79.3
Skywork-Reward-V2-Llama-3.2-3B2025.07		64.3	65.3	87.5	59.5	69.2
o3-mini2025.07		63	69.4	83.4	83.3	74.8
Claude-3.5-Sonnet2025.07		62.3	66.3	66.1	64.3	64.8
o3-mini2025.07		62.3	86.7	85.7	92.9	81.9
Skywork-Reward-V2-Qwen3-0.6B2025.07		62.3	66.3	82.1	59.5	67.5
Skywork-Reward-V2-Llama-3.2-1B2025.07		61	66.3	73.2	59.5	65
DeepSeek-R12025.07		59.1	82.7	80.4	92.9	78.8
o3-mini2025.07		58.4	62.2	82.1	78.6	70.3
GPT-4o2025.07		50.6	54.1	75	59.5	59.8