Open Arabic LLM - Arabic MMLU HT Professional LAW: leaderboard
Metric: Accuracy (%). Source: huggingface.co. 163 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 2.5 72B Instruct | 67.84 |
| 2 | Qwen 2 72B | 64.33 |
| 3 | Llama 3.3 70B Instruct | 64.18 |
| 4 | Qwen 2.5 72B | 63.73 |
| 5 | Qwentile2.5-32B-Instruct | 61.17 |
| 6 | Awqward2.5-32B-Instruct | 61.01 |
| 7 | gemma-3-27B-pt | 60.85 |
| 8 | Qwen 2.5 32B Instruct | 60.76 |
| 9 | Qwen2.5-32B-Instruct-CFT | 60.66 |
| 10 | Gemma 3 27B (IT) | 60.19 |
| 11 | Qwen2.5-32B-Instruct-abliterated-v2 | 59.41 |
| 12 | free-evo-qwen72B-v0.8-re | 58.42 |
| 13 | Saka-14B | 57.61 |
| 14 | aya-expanse-32B | 55.98 |
| 15 | gemma-3-12B-pt | 55.63 |
Interactive version: theaggregate.ai/benchmark?slug=open-arabic-llm-arabic-mmlu-ht-professional-law · How It Works · Data refreshed daily, snapshot 2026-09-05.