Long-term state poisoning evaluation on OpenClaw Log Replay
4.34Harm Score (HS)Kimi k2.5
Evaluation Results
| Method | Links | |
|---|---|---|
| Kimi k2.52026.05 | 4.34 | |
| Grok-12026.05 | 4.8 | |
| GPT-4o2026.05 | 5.03 | |
| MiniMax-abab6.52026.05 | 5.73 |
| Method | Links | |
|---|---|---|
| Kimi k2.52026.05 | 4.34 | |
| Grok-12026.05 | 4.8 | |
| GPT-4o2026.05 | 5.03 | |
| MiniMax-abab6.52026.05 | 5.73 |