ResearchBenchmarksSycophancy Evaluation on Sycophancy Evaluation Control conditionFollow2.64Control MeanGen 2.51.87042.07022.272.4698Apr 19, 2026Evaluation ResultsMethodMethodLinksControl Mean95% CI Lower Bound (Control)Gen 2.5Generation=2.5, Guardr...Generation=2.5, Guardrails=None (Control condition)2026.042.642.56Gen 3.0Generation=3.0, Guardr...Generation=3.0, Guardrails=None (Control condition)2026.042.011.94Gen 2.0Generation=2.0, Guardr...Generation=2.0, Guardrails=None (Control condition)2026.041.91.82