Open Arabic LLM - Arabic MMLU Arabic Language (Primary School): leaderboard
The "Arabic MMLU Arabic Language (Primary School)" subtask of Open Arabic LLM. A task from the Open Arabic LLM Leaderboard, evaluating models on Arabic-language knowledge, reasoning and understanding across native and translated benchmarks.
Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 163 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | ALLaM-7B-Instruct-preview | 81.75 | #718 |
| 2 | Yehia-7B-preview | 79.76 | #646 |
| 3 | Awqward2.5-32B-Instruct | 77.78 | #371 |
| 4 | Qwen 2.5 32B Instruct | 76.98 | #498 |
| 5 | Qwen2.5-32B-Instruct-CFT | 76.98 | #375 |
| 6 | Qwen2.5-32B-Instruct-abliterated-v2 | 76.19 | #394 |
| 7 | Qwen 2.5 14B | 75.79 | #725 |
| 8 | AIC-1 | 75.79 | #231 |
| 9 | Saka-14B | 75.4 | #344 |
| 10 | free-evo-qwen72B-v0.8-re | 75.4 | #435 |
| 11 | Qwentile2.5-32B-Instruct | 75 | #339 |
| 12 | EZO-Qwen2.5-32B-Instruct | 74.6 | #471 |
| 13 | Qwen3-8B-Base | 74.21 | #626 |
| 14 | tempmotacilla-cinerea-0308 | 74.21 | #395 |
| 15 | Gemma 3 27B (IT) | 72.62 | #523 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=open-arabic-llm-arabic-mmlu-arabic-language-primary-school · How It Works · Data refreshed daily, snapshot 2026-10-20.