Deception Resistance (Lechmazur): leaderboard
Measures LLM resilience to misleading arguments: models must maintain accurate reasoning when exposed to persuasive disinformation from other models.
Metric: Vulnerability Score (lower is better). Source: github.com. Saturation forecast: Estimated already saturated. 18 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude 3 Opus | 0.28 | #571 |
| 2 | Claude 3.5 Sonnet | 0.28 | #350 |
| 3 | O1 Preview | 0.32 | #193 |
| 4 | Mistral Large 2 (Jul) | 0.35 | #484 |
| 5 | O1 Mini | 0.5 | #323 |
| 6 | Llama 3.1 405B | 0.54 | #466 |
| 7 | Qwen 2.5 72B | 0.61 | #499 |
| 8 | GPT-4o | 0.61 | #343 |
| 9 | Gemini 1.5 Pro (Sept) | 0.62 | #448 |
| 10 | Gemini 1.5 Flash | 0.64 | #618 |
| 11 | Claude 3 Haiku | 0.78 | #803 |
| 12 | Grok 2 | 0.81 | #452 |
| 13 | Llama 3.1 70B | 0.82 | #592 |
| 14 | GPT-4o Mini | 1.07 | #599 |
| 15 | GPT-4 Turbo | 1.18 | #404 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=deception-resistance-lechmazur · How It Works · Data refreshed daily, snapshot 2026-10-20.