Open Arabic LLM - Arabic MMLU Civics (High School): leaderboard
The "Arabic MMLU Civics (High 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 2030. 163 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 2.5 72B Instruct | 55.17 | #454 |
| 2 | AceGPT-v2-32B-Chat | 55.17 | #507 |
| 3 | Qwen 2.5 32B Instruct | 54.02 | #498 |
| 4 | Awqward2.5-32B-Instruct | 54.02 | #371 |
| 5 | Saka-14B | 54.02 | #344 |
| 6 | Qwen2.5-32B-Instruct-CFT | 54.02 | #375 |
| 7 | Llama 3.3 70B Instruct | 52.87 | #534 |
| 8 | AceGPT-v2-70B-Chat | 52.87 | #407 |
| 9 | EZO-Qwen2.5-32B-Instruct | 52.87 | #471 |
| 10 | Qwen 2.5 14B | 51.72 | #725 |
| 11 | Fanar-1-9B-Instruct | 51.72 | #633 |
| 12 | Qwen 2.5 3B | 50.57 | #1211 |
| 13 | Qwen 2.5 72B | 50.57 | #499 |
| 14 | Qwentile2.5-32B-Instruct | 50.57 | #339 |
| 15 | aya-23-35B | 49.43 | #883 |
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-high-school · How It Works · Data refreshed daily, snapshot 2026-10-20.