HELM AIR-Bench 2024 - #39-40.52: Classification of individuals - Physical Characteristics: leaderboard
HELM AIR-Bench 2024 (v1.19.0) refusal rate for level-4 risk category "#39-40.52: Classification of individuals - Physical Characteristics" (level-3 "#39-40: Discrimination/Protected Characteristics Combinations", level-2 "Discrimination/Bias"). Only 27 prompts, so scores move in coarse steps. A GPT-4o judge scores each response from compliance (0) to refusal (1); shown as 0-100, higher means more refusals (safer).
Metric: Refusal Rate (%). Source: crfm.stanford.edu. Saturation forecast: Estimated already saturated. 87 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 100 | #105 |
| 2 | Claude Sonnet 4.5 | 100 | #148 |
| 3 | GPT-5 Nano | 100 | #439 |
| 4 | Claude Haiku 4.5 (20251001) | 100 | #264 |
| 5 | O3 (2025-04-16) | 100 | #125 |
| 6 | Claude 3 Opus (20240229) | 100 | #428 |
| 7 | GPT-5.1 (2025-11-13) | 100 | #107 |
| 8 | Claude Sonnet 4 (20250514) | 100 | #223 |
| 9 | Claude 3.5 Sonnet (20241022) | 96.3 | #302 |
| 10 | Grok 4 | 96.3 | #189 |
| 11 | Claude 3.5 Sonnet (20240620) | 96.3 | #308 |
| 12 | Qwen 3 Next 80B A3B (Thinking) | 96.3 | #312 |
| 13 | Claude 3.7 Sonnet (20250219) | 96.3 | #209 |
| 14 | GPT-OSS-120B | 92.59 | #346 |
| 15 | Gemini 1.5 Flash (002) | 92.59 | #587 |
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-2024-39-40-52-classification-of-individuals-physical-characteristics · How It Works · Data refreshed daily, snapshot 2026-11-05.