Share your thoughts, 1 month free Claude Pro on usSee more

Verifiable Judging on VerifyBench Hard

88.8Accuracy

OpenAI/GPT-o1

Updated 1mo ago

Evaluation Results

Method	Links
OpenAI/GPT-o1 2025.07		88.8	85.48
Master-RM-32B 2025.07		86.8	81.96
Anthropic/Claude-4-Sonnet 2025.07		85.3	79.71
Master-RM-7B 2025.07		84.4	80.98
OpenAI/GPT-4o 2025.07		84.3	77.94
OpenAI/GPT-4o-mini 2025.07		82.8	76.29
Multi-sub RM 2025.07		82.5	78.42
Qwen2.5-32B-Instruct 2025.07		81.3	75.3
Qwen2.5-7B-Instruct 2025.07		80.2	74.21
Qwen2.5-14B-Instruct 2025.07		78.4	71.79
Qwen2.5-72B-Instruct 2025.07		78.3	72.63
Llama-3-70B-Instruct 2025.07		77.1	73.35
math-verify 2025.07		76	60.21
Qwen2.5-3B-Instruct 2025.07		75.2	72.56
Omni-Judge 2025.07		67.7	58.98
Qwen2.5-1.5B-Instruct 2025.07		67.7	66.7
Llama-3-8B-Instruct 2025.07		61.3	60.56
General-Verifier 2025.07		50.2	49.4
Qwen2.5-0.5B-Instruct 2025.07		40.7	40.55