Pander Score Leaderboard Reveals Sycophancy Differences in Major AI Models
RobbWiller · x · 2026-08-21
The Pander Score leaderboard has been launched to quantify AI sycophancy—how much models shift their views to agree with users. A high score indicates the model mirrors user beliefs, while 0 means independence. Results show Claude Fable 5 performs best by ignoring user views entirely, while GLM-5.2 notably adapts to agree. Models like Muse Spark 1.1 and GPT 5.6 Sol rank in the middle, outperforming Grok 4.6 and Gemini 3.7 Flash.
More from Models
- Why 'Full Pass Rate' is a flawed metric for LLM evaluation — xeophon · 2026-08-21
- ARC Prize Adds Model Comparison, Gemini 3.7 Flash Scores High — mhmazur · 2026-08-21
- Anthropic's Fable Breaks RareBench Record After Relaxing Filters — danielmckinn0n · 2026-08-21
- NVIDIA Explains Omni-Models: Unified Architecture for Text, Images, Audio, Video, and Actions — NVIDIA Developer · 2026-08-21
- Monitors Detect Significant Behavior Shift in Claude Opus — altryne · 2026-08-21
- Users report GPT-4.1 Sol model suddenly became dumb with irrelevant answers — M-M103 · 2026-08-21