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

PRBench

Benchmarks

Task NameDataset NameSOTA ResultTrend
Legal ReasoningPRBench Legal-Hard
GPT-5 Grader Score0.377
4
Legal ReasoningPRBench Legal
GPT-5 Grader Score0.482
4
Legal reasoningPRBench-Legal 500 prompts
GPT-5 Grader Score44.1
4
Generative PerformancePRBench
Pearson r0.8
1
Showing 4 of 4 rows