Open Arabic LLM - Arabic MMLU Computer Science (University): leaderboard
The "Arabic MMLU Computer Science (University)" 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 | Qwen2.5-32B-Instruct-CFT | 82.81 | #375 |
| 2 | lambda-qwen2.5-14B-dpo-test | 81.25 | #481 |
| 3 | Awqward2.5-32B-Instruct | 81.25 | #371 |
| 4 | Qwen 2.5 32B Instruct | 79.69 | #498 |
| 5 | Qwen 2.5 72B Instruct | 78.12 | #454 |
| 6 | Qwen2.5-32B-Instruct-abliterated-v2 | 78.12 | #394 |
| 7 | Qwentile2.5-32B-Instruct | 78.12 | #339 |
| 8 | Saka-14B | 78.12 | #344 |
| 9 | Qwen2.5-Lumen-14B | 78.12 | #477 |
| 10 | Qwen3-8B-Base | 76.56 | #626 |
| 11 | AceGPT-v2-32B-Chat | 76.56 | #507 |
| 12 | Fanar-1-9B-Instruct | 76.56 | #633 |
| 13 | EZO-Qwen2.5-32B-Instruct | 76.56 | #471 |
| 14 | free-evo-qwen72B-v0.8-re | 76.56 | #435 |
| 15 | AIC-1 | 76.56 | #231 |
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-university · How It Works · Data refreshed daily, snapshot 2026-10-20.