Open Arabic LLM - Aratrust Trustfulness — leaderboard
Metric: Accuracy (%). Source: huggingface.co. 162 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Llama 3.3 70B Instruct | 82.05 |
| 2 | Gemma 3 27B (IT) | 80.77 |
| 3 | Qwen 2.5 32B Instruct | 80.77 |
| 4 | Qwen 2.5 72B Instruct | 79.49 |
| 5 | gemma-3-27B-pt | 79.49 |
| 6 | aya-expanse-32B | 79.49 |
| 7 | Qwen 2.5 72B | 78.21 |
| 8 | Qwen2.5-32B-Instruct-abliterated-v2 | 78.21 |
| 9 | Awqward2.5-32B-Instruct | 78.21 |
| 10 | calme-2.2-qwen2.5-72B | 78.21 |
| 11 | lambda-qwen2.5-32B-dpo-test | 78.21 |
| 12 | Saka-14B | 76.92 |
| 13 | Qwen2.5-32B-Instruct-CFT | 76.92 |
| 14 | EZO-Qwen2.5-32B-Instruct | 76.92 |
| 15 | calme-2.1-qwen2.5-72B | 76.92 |
Interactive version: theaggregate.ai/benchmark?slug=open-arabic-llm-aratrust-trustfulness · How the rankings work · Data refreshed daily, snapshot 2026-07-22.