Nejumi 4 - Toxicity - Social Norms: leaderboard
Metric: Criteria met (%). Source: nejumi.ai. 137 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5 Nano (High) | 99.6 |
| 2 | GPT-5.4 (High) | 99.6 |
| 3 | GPT-5.2 (xHigh) | 99.6 |
| 4 | Claude Opus 4.8 (xHigh) | 99.6 |
| 5 | Qwen 3.6 27B (Thinking) | 99.6 |
| 6 | GPT-5.5 (xHigh) | 99.21 |
| 7 | GPT-5.4 Mini (High) | 99.21 |
| 8 | DeepSeek V3.2 (Thinking) | 99.21 |
| 9 | Claude Sonnet 4 (Non-reasoning) | 99.21 |
| 10 | GPT-5.6 Luna (Max) | 98.81 |
| 11 | ELYZA-Shortcut-1.0-Qwen-32B | 98.81 |
| 12 | K-EXAONE-236B-A23B (Thinking) | 98.81 |
| 13 | Qwen 3.6 Max (Preview) (Thinking) | 98.81 |
| 14 | GPT-5.6 Terra (Max) | 98.41 |
| 15 | Claude Sonnet 4.6 (Thinking) | 98.41 |
Interactive version: theaggregate.ai/benchmark?slug=nejumi-4-toxicity-social-norms · How It Works · Data refreshed daily, snapshot 2026-09-19.