HELM AIR-Bench 2024 - #39-40.25: Characterization of identity - Sexual orientation: leaderboard
HELM AIR-Bench 2024 (v1.19.0) refusal rate for level-4 risk category "#39-40.25: Characterization of identity - Sexual orientation" (level-3 "#39-40: Discrimination/Protected Characteristics Combinations", level-2 "Discrimination/Bias"). Only 30 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: Around February 2027. 87 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 Next 80B A3B (Thinking) | 83.33 | #312 |
| 2 | GPT-5.1 (2025-11-13) | 80 | #107 |
| 3 | GPT-5 Mini (2025-08-07) | 78.33 | #175 |
| 4 | GPT-5 | 75 | #105 |
| 5 | Claude Haiku 4.5 (20251001) | 71.67 | #264 |
| 6 | O3 Mini (2025-01-31) | 70 | #249 |
| 7 | Gemini 2.5 Flash (Preview 04-17) | 70 | #246 |
| 8 | Claude Sonnet 4.5 | 68.33 | #148 |
| 9 | GPT-5 Nano | 68.33 | #439 |
| 10 | Grok 3 Mini Beta | 68.33 | #305 |
| 11 | Kimi K2 | 63.33 | #252 |
| 12 | DeepSeek R1 | 61.67 | #262 |
| 13 | GPT-4.1 (2025-04-14) | 61.67 | #257 |
| 14 | Grok 4 | 60 | #189 |
| 15 | Gemini 2.5 Pro (Preview 03-25) | 60 | #126 |
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-2024-39-40-25-characterization-of-identity-sexual-orientation · How It Works · Data refreshed daily, snapshot 2026-11-05.