Open Arabic LLM - Arabic MMLU Driving Test: leaderboard
Metric: Accuracy (%). Source: huggingface.co. 163 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 2.5 72B Instruct | 77.21 |
| 2 | Qwen 2 72B | 76.8 |
| 3 | gemma-3-27B-pt | 76.14 |
| 4 | Saka-14B | 75.89 |
| 5 | Gemma 3 27B (IT) | 75.47 |
| 6 | Qwentile2.5-32B-Instruct | 74.81 |
| 7 | tempmotacilla-cinerea-0308 | 73.82 |
| 8 | Qwen3-8B-Base | 73.33 |
| 9 | Qwen2.5-32B-Instruct-abliterated-v2 | 73.33 |
| 10 | free-evo-qwen72B-v0.8-re | 73.25 |
| 11 | Qwen 2.5 14B | 73 |
| 12 | lambda-qwen2.5-14B-dpo-test | 72.42 |
| 13 | Llama 3.3 70B Instruct | 72.25 |
| 14 | Qwen 2.5 72B | 72.01 |
| 15 | Qwen2.5-32B-Instruct-CFT | 71.59 |
Interactive version: theaggregate.ai/benchmark?slug=open-arabic-llm-arabic-mmlu-driving-test · How It Works · Data refreshed daily, snapshot 2026-09-05.