XL-SafetyBench: leaderboard
Country-grounded cross-cultural safety benchmark with 5,500 test cases across 10 country-language pairs, separating universal jailbreak robustness from culturally embedded sensitivities.
Metric: Overall ASR (self-reported). Source: arxiv.org. Source: benchmarklist.com. Saturation forecast: Around June 2027. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Sonnet 4.5 | 2.8 | #142 |
| 2 | Claude Opus 4.6 | 5.9 | #63 |
| 3 | Qwen 3.5 397B A17B | 18.1 | #146 |
| 4 | Grok 4.20 | 30.6 | #175 |
| 5 | Gemini 3.1 Pro (Preview) | 43.4 | #58 |
| 6 | GPT-5.4 | 47.1 | #79 |
| 7 | Gemini 3 Flash | 50 | #99 |
| 8 | GPT-5 Mini | 59.2 | #186 |
| 9 | Llama 4 Maverick | 92 | #464 |
| 10 | Mistral Large 3 | 98.8 | #401 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=xl-safetybench · How It Works · Data refreshed daily, snapshot 2026-10-20.