Open Arabic LLM - Arabic MMLU Computer Science (High School): leaderboard
The "Arabic MMLU Computer Science (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 January 2027. 163 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 2.5 72B Instruct | 80.08 | #454 |
| 2 | AIC-1 | 74.71 | #231 |
| 3 | Qwen2.5-32B-Instruct-abliterated-v2 | 72.41 | #394 |
| 4 | Awqward2.5-32B-Instruct | 72.03 | #371 |
| 5 | Qwentile2.5-32B-Instruct | 71.65 | #339 |
| 6 | EZO-Qwen2.5-32B-Instruct | 71.65 | #471 |
| 7 | Qwen 2.5 32B Instruct | 70.88 | #498 |
| 8 | Qwen2.5-32B-Instruct-CFT | 70.88 | #375 |
| 9 | Qwen 2 72B | 70.5 | #521 |
| 10 | Saka-14B | 70.5 | #344 |
| 11 | free-evo-qwen72B-v0.8-re | 70.5 | #435 |
| 12 | Llama 3.3 70B Instruct | 70.11 | #534 |
| 13 | Gemma 3 27B (IT) | 68.97 | #523 |
| 14 | AceGPT-v2-70B-Chat | 68.97 | #407 |
| 15 | Qwen 2.5 72B | 68.58 | #499 |
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-high-school · How It Works · Data refreshed daily, snapshot 2026-10-20.