ResearchBenchmarksInstruction Following on IFEval (CSR, TCR, Conf. Res., Avg-T)Follow88.7CSR (Correctness Score Rate)CRGC78.50881.15483.886.446Jun 2, 2026Evaluation ResultsMethodMethodLinksCSR (Correctness Score Rate)TCR (Task Completion Rate)Conf. Res. (Confidence Resolution)Avg-T (Average Time)CRGCBackbone=GPT-4oBackbone=GPT-4o2026.0688.789.684.61.15DeCRIMBackbone=GPT-4oBackbone=GPT-4o2026.0686.287.568.43.85SPBackbone=GPT-4oBackbone=GPT-4o2026.0678.980.341.21