Open Arabic LLM - Arabic MMLU Civics (Middle School): leaderboard
The "Arabic MMLU Civics (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 | Qwen 2.5 72B Instruct | 63.56 | #454 |
| 2 | gemma-3-27B-pt | 63.14 | #502 |
| 3 | AceGPT-v2-32B-Chat | 61.86 | #507 |
| 4 | Qwen 2.5 14B | 60.59 | #725 |
| 5 | lambda-qwen2.5-14B-dpo-test | 60.17 | #481 |
| 6 | Saka-14B | 60.17 | #344 |
| 7 | Qwen2.5-32B-Instruct-abliterated-v2 | 59.75 | #394 |
| 8 | ALLaM-7B-Instruct-preview | 59.32 | #718 |
| 9 | free-evo-qwen72B-v0.8-re | 58.9 | #435 |
| 10 | AceGPT-v2-70B-Chat | 58.47 | #407 |
| 11 | Yehia-7B-preview | 58.47 | #646 |
| 12 | tempmotacilla-cinerea-0308 | 58.05 | #395 |
| 13 | AIC-1 | 58.05 | #231 |
| 14 | Llama 3.3 70B Instruct | 57.63 | #534 |
| 15 | Hala-9B | 57.63 | #504 |
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-civics-middle-school · How It Works · Data refreshed daily, snapshot 2026-10-20.