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

Verifiable Judging on VerifyBench (full)

95.7Accuracy

OpenAI/GPT-o1

Updated 1mo ago

Evaluation Results

Method	Links
OpenAI/GPT-o1 2025.07		95.7	95.7
Master-RM-32B 2025.07		95.15	95.14
Anthropic/Claude-4-Sonnet 2025.07		95	95
Multi-sub RM 2025.07		95	95
Master-RM-7B 2025.07		94.45	94.45
Qwen2.5-72B-Instruct 2025.07		94.3	94.3
OpenAI/GPT-4o 2025.07		94.15	94.15
Qwen2.5-32B-Instruct 2025.07		93.25	93.25
Llama-3-70B-Instruct 2025.07		92.5	92.49
Qwen2.5-14B-Instruct 2025.07		92.4	92.4
OpenAI/GPT-4o-mini 2025.07		91.4	91.37
Qwen2.5-7B-Instruct 2025.07		89.05	89
Qwen2.5-3B-Instruct 2025.07		88.35	88.35
Qwen2.5-1.5B-Instruct 2025.07		83.95	83.88
Omni-Judge 2025.07		80.2	80.03
Llama-3-8B-Instruct 2025.07		79.95	79.8
General-Verifier 2025.07		67.65	67.46
math-verify 2025.07		66.95	63.4
Qwen2.5-0.5B-Instruct 2025.07		55.15	49.54