List your company

Sycophancy Resistance on Anthropic model-written-evals sycophancy suite

94.6Overall Sycophancy Resistance ScoreInstruction Tuned
90.54491.59792.6593.703Jun 4, 2026

Evaluation results

MethodLinks
Instruction Tuned2026.0694.687.4
BCT2026.0694.387.7
Attention Consistency Training (AttCT)2026.0694.285.6
ACT2026.0694.185.3
MLP Consistency Training (MLPCT)2026.0693.884.4
Baseline2026.0690.776