Open Arabic LLM - Arabic MMLU Arabic Language (Middle School): leaderboard
The "Arabic MMLU Arabic Language (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: Around January 2027. 163 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | lambda-qwen2.5-14B-dpo-test | 85.19 | #481 |
| 2 | AceGPT-v2-32B-Chat | 81.48 | #507 |
| 3 | tempmotacilla-cinerea-0308 | 81.48 | #395 |
| 4 | Saka-14B | 81.48 | #344 |
| 5 | Llama 3.3 70B Instruct | 77.78 | #534 |
| 6 | Qwen 2.5 32B Instruct | 77.78 | #498 |
| 7 | gemma-3-27B-pt | 77.78 | #502 |
| 8 | Qwen 2.5 14B | 77.78 | #725 |
| 9 | jais-adapted-70B-chat | 77.78 | #636 |
| 10 | Qwen 2 72B | 77.78 | #521 |
| 11 | Awqward2.5-32B-Instruct | 77.78 | #371 |
| 12 | Qwen2.5-32B-Instruct-CFT | 77.78 | #375 |
| 13 | Gemma 3 27B (IT) | 74.07 | #523 |
| 14 | Qwen 2.5 72B Instruct | 74.07 | #454 |
| 15 | Qwen2.5-32B-Instruct-abliterated-v2 | 74.07 | #394 |
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-middle-school · How It Works · Data refreshed daily, snapshot 2026-10-20.