Open Arabic LLM - Arabic MMLU Islamic Studies: leaderboard
The "Arabic MMLU Islamic Studies" 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 | Qwen 2.5 72B Instruct | 67.29 | #454 |
| 2 | ALLaM-7B-Instruct-preview | 66.2 | #718 |
| 3 | Yehia-7B-preview | 66.2 | #646 |
| 4 | Llama 3.3 70B Instruct | 65.73 | #534 |
| 5 | AceGPT-v2-70B-Chat | 63.85 | #407 |
| 6 | jais-adapted-70B-chat | 62.91 | #636 |
| 7 | gemma-3-27B-pt | 62.28 | #502 |
| 8 | aya-expanse-32B | 62.13 | #692 |
| 9 | Gemma 3 27B (IT) | 58.22 | #523 |
| 10 | EZO-Qwen2.5-32B-Instruct | 57.9 | #471 |
| 11 | Qwen 2.5 32B Instruct | 56.96 | #498 |
| 12 | jais-family-30B-16k-chat | 56.96 | #755 |
| 13 | Awqward2.5-32B-Instruct | 56.96 | #371 |
| 14 | Qwen2.5-32B-Instruct-abliterated-v2 | 56.81 | #394 |
| 15 | Qwentile2.5-32B-Instruct | 56.81 | #339 |
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-islamic-studies · How It Works · Data refreshed daily, snapshot 2026-10-20.