Sycophancy Evaluation on MIST Moral
1.6SycophancySonnet 4.6
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Sonnet 4.6Memory System=Chat History2026.06 | 1.6 | — | — | |
| GPT-5.2Memory System=Chat History2026.06 | 6.1 | — | — | |
| Qwen 3.5Memory System=Chat History2026.06 | 6.4 | — | — | |
| MiniMax 2.5Memory System=Chat History2026.06 | 9.1 | — | — | |
| SummarizationStrategy=Summarize2026.06 | 12.8 | 83 | 12.8 | |
| Sonnet 4.6Memory System=Zep2026.06 | 15.9 | — | — | |
| Kimi K2.5Memory System=Chat History2026.06 | 16.2 | — | — | |
| GPT-5.2Memory System=Zep2026.06 | 17.1 | — | — | |
| MiniMax 2.5Memory System=Zep2026.06 | 17.3 | — | — | |
| Assistant Role InclusionStrategy=Asst. Role2026.06 | 20.3 | 76 | 20.5 | |
| Qwen 3.5Memory System=Zep2026.06 | 21 | — | — | |
| Sonnet 4.6Memory System=MemOS2026.06 | 22.3 | — | — | |
| Anti-Sycophancy PromptingStrategy=Prompting2026.06 | 26.5 | 69.5 | 26.5 | |
| MiniMax 2.5Memory System=MemOS2026.06 | 30.6 | — | — | |
| Kimi K2.5Memory System=Zep2026.06 | 31.8 | — | — | |
| GPT-5.2Memory System=MemOS2026.06 | 34.6 | — | — | |
| Sonnet 4.6Memory System=Mem02026.06 | 40.2 | — | — | |
| Mem0Strategy=Baseline2026.06 | 41 | 56 | 41.2 | |
| GPT-5.2Memory System=Mem02026.06 | 41 | — | — | |
| MiniMax 2.5Memory System=Mem02026.06 | 46.7 | — | — | |
| Qwen 3.5Memory System=MemOS2026.06 | 47.9 | — | — | |
| Qwen 3.5Memory System=Mem02026.06 | 55.3 | — | — | |
| Kimi K2.5Memory System=MemOS2026.06 | 61.1 | — | — | |
| Kimi K2.5Memory System=Mem02026.06 | 69.8 | — | — |