HELM AIR-Bench 2024 - #18: Erotic: leaderboard
HELM AIR-Bench 2024 (release v1.19.0) refusal rate for the level-3 risk category "#18: Erotic", under the level-2 category Sexual Content of the AIR 2024 regulation-derived taxonomy. Every prompt is a harmful request; a GPT-4o judge grades each response from full compliance (0) to refusal (1), and the mean is shown as 0-100, where higher means more refusals (safer).
Metric: Refusal Rate (%). Source: crfm.stanford.edu. Saturation forecast: Not forecast. 87 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude 3.5 Sonnet (20241022) | 95 | #302 |
| 2 | Claude Sonnet 4 (20250514) | 91.67 | #223 |
| 3 | Claude Opus 4 (20250514) | 86.67 | #164 |
| 4 | Claude 3.5 Sonnet (20240620) | 85 | #308 |
| 5 | Claude Haiku 4.5 (20251001) | 81.67 | #264 |
| 6 | Claude 3 Sonnet (20240229) | 80 | #694 |
| 7 | Llama 3 8B Instruct | 75 | #1116 |
| 8 | Claude 3.7 Sonnet (20250219) | 75 | #209 |
| 9 | GPT-5.1 (2025-11-13) | 71.67 | #107 |
| 10 | GPT-OSS-20B | 68.33 | #523 |
| 11 | Claude 3 Haiku (20240307) | 68.33 | #780 |
| 12 | GPT-OSS-120B | 66.67 | #346 |
| 13 | Claude Sonnet 4.5 | 63.33 | #148 |
| 14 | Qwen 3 Next 80B A3B (Thinking) | 60 | #312 |
| 15 | Claude 3 Opus (20240229) | 58.33 | #428 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=helm-air-bench-2024-18-erotic · How It Works · Data refreshed daily, snapshot 2026-11-05.