LLM Persuasion Average Strength
Chat & Writing · 2026-03-26
Measures how much a model persuader moves an opponent model toward its assigned side in multi-turn persuasion dialogues. Higher scores indicate stronger average persuasive effect.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-5.4 | 1.7 |
| Opus 4.6 | 1.7 |
| Doubao Seed 2.0 Pro | 1.6 |
| Sonnet 4.6 | 1.6 |
| Kimi K2.5 | 1.3 |
| Gemini 3.1 Pro Preview | 1.2 |
| GLM-5 | 1.2 |
| Grok 4.20 | 1.2 |
| Qwen3.5 397B A17B | 1.0 |
| MiniMax M2.7 | 1.0 |
| Gemini 3.1 Flash-Lite Preview | 0.8 |
| ERNIE 5.0 | 0.8 |
| DeepSeek-V3.2 | 0.7 |
| MiMo-V2-Pro | 0.5 |
| Mistral Large 3 675B Instruct 2512 | 0.4 |