Enkrypt AI - Toxicity Risk: leaderboard
Enkrypt AI’s LLM safety leaderboard: the rate of toxic output under provocation. It is reported as a risk, so lower is better here and the leader is the least risky model rather than the most capable one.
Metric: Risk Score. Source: leaderboard.enkryptai.com. Saturation forecast: Estimated already saturated. 298 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Llama 3.1 8B Instruct | 0 | #1035 |
| 2 | Qwen 2.5 72B Instruct | 0 | #454 |
| 3 | Claude 3.5 Sonnet (20240620) | 0 | #299 |
| 4 | Llama 3 70B Instruct | 0 | #634 |
| 5 | Gemma 2 27B (IT) | 0 | #645 |
| 6 | Claude Fable 5 | 0 | #16 |
| 7 | Mistral-NeMo-Minitron-8B-Instruct | 0 | #907 |
| 8 | Smaug-Llama-3-70B-Instruct | 0 | #716 |
| 9 | Claude 3.5 Sonnet (20241022) | 0.09 | #291 |
| 10 | Claude 3 Opus (20240229) | 0.09 | #412 |
| 11 | granite-3.0-3B-a800m-instruct | 0.09 | #1418 |
| 12 | Claude Sonnet 5 | 0.18 | #57 |
| 13 | granite-3.1-2B-instruct | 0.18 | #1385 |
| 14 | Claude Sonnet 4 (20250514) | 0.23 | #213 |
| 15 | granite-3.1-3B-a800m-instruct | 0.23 | #1452 |
No result here: #6 GPT-6 Astra, #7 GPT-6.1 Sol, #8 Gemini 4 Argon.
Interactive version: theaggregate.ai/benchmark?slug=enkrypt-ai-toxicity-risk · How It Works · Data refreshed daily, snapshot 2026-10-20.