Arabic Broad Leaderboard - Trust & Safety: leaderboard
Arabic trust and safety prompts: the 30-question Trust & Safety category of SILMA's Arabic Broad Benchmark, whose answers are scored 0-10 by rules or an LLM judge. Category average score (0-10), higher is better; each model's latest result file is used.
Metric: Average Score (0-10). Source: huggingface.co. Saturation forecast: Estimated already saturated. 121 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 | 10 | #33 |
| 2 | GLM-5 | 10 | #149 |
| 3 | Claude Sonnet 4 (20250514) | 10 | #223 |
| 4 | Muse Spark 1.3 | 10 | #31 |
| 5 | Muse Spark 1.1 | 10 | #55 |
| 6 | GPT-4o | 9.67 | #357 |
| 7 | Gemini 3.1 Pro (Preview) | 9.67 | #64 |
| 8 | Gemini 3.8 Flash | 9.67 | #30 |
| 9 | DeepSeek V4.1 Flash | 9.67 | #37 |
| 10 | Claude Sonnet 5.5 | 9.67 | #14 |
| 11 | Muse Glimmer 30B | 9.67 | #191 |
| 12 | Qwen 3.6 Max Preview | 9.67 | #91 |
| 13 | Qwen3.8 2.4T A95B | 9.67 | #59 |
| 14 | GPT-6.1 Pro Sol | 9.67 | #11 |
| 15 | GPT-5 | 9.33 | #105 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=arabic-broad-leaderboard-trust-safety · How It Works · Data refreshed daily, snapshot 2026-11-05.