ResearchBenchmarksConciseness Correlation on WikiEvalConciseness Correlation on WikiEvalFollow0.628Spearman Correlation (rs)ConCISE0.08720.22760.3680.5084Nov 20, 2025Evaluation resultsMethodMethodLinksSpearman Correlation (rs)Spearman p-value (ps)Kendall's Tau (tau)Kendall p-value (pk)ConCISELLM Judge Model=GPT-4oLLM Judge Model=GPT-4o2025.110.6280.0010.5230.001ConCISELLM Judge Model=Claude...LLM Judge Model=Claude-4Sonnet2025.110.5370.0010.4360.001ConCISELLM Judge Model=Gemini...LLM Judge Model=Gemini-2.0-Flash2025.110.5180.0010.4220.001ConCISELLM Judge Model=Mistra...LLM Judge Model=Mistral-Large-22025.110.4730.00170.3760.002GPT Score2025.110.1080.4740.0870.49