Open Arabic LLM - Arabic MMLU Computer Science (Middle School): leaderboard
The "Arabic MMLU Computer Science (Middle 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 | Qwen3-8B-Base | 96.3 | #626 |
| 2 | Qwen 2.5 14B | 96.3 | #725 |
| 3 | ALLaM-7B-Instruct-preview | 96.3 | #718 |
| 4 | lambda-qwen2.5-14B-dpo-test | 96.3 | #481 |
| 5 | Saka-14B | 96.3 | #344 |
| 6 | Qwen 2.5 14B Instruct | 92.59 | #647 |
| 7 | Qwen 2 7B Instruct | 92.59 | #991 |
| 8 | Qwen 2 72B | 92.59 | #521 |
| 9 | Qwen2.5-32B-Instruct-abliterated-v2 | 92.59 | #394 |
| 10 | Qwentile2.5-32B-Instruct | 92.59 | #339 |
| 11 | tempmotacilla-cinerea-0308 | 92.59 | #395 |
| 12 | Yehia-7B-preview | 92.59 | #646 |
| 13 | c4ai-command-r7B-arabic-02-2025 | 92.59 | #732 |
| 14 | Fanar-1-9B | 92.59 | #778 |
| 15 | Qwen2.5-14B-Gutenberg-1e-Delta | 92.59 | #473 |
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-computer-science-middle-school · How It Works · Data refreshed daily, snapshot 2026-10-20.