Open Arabic LLM - Arabic MMLU HT Abstract Algebra: leaderboard
The "Arabic MMLU HT Abstract Algebra" 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 2033. 163 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 2.5 72B Instruct | 47 | #454 |
| 2 | AIC-1 | 47 | #231 |
| 3 | Qwen 2.5 72B | 43 | #499 |
| 4 | Qwen 2.5 14B | 41 | #725 |
| 5 | Qwen3-8B-Base | 40 | #626 |
| 6 | Awqward2.5-32B-Instruct | 40 | #371 |
| 7 | aya-expanse-32B | 39 | #692 |
| 8 | Hala-9B | 39 | #504 |
| 9 | AceGPT-v2-32B-Chat | 38 | #507 |
| 10 | Qwen3-4B-Base | 38 | #897 |
| 11 | Qwen2.5-32B-Instruct-abliterated-v2 | 38 | #394 |
| 12 | Saka-14B | 38 | #344 |
| 13 | Qwen3-14B-Base | 37 | #623 |
| 14 | Qwentile2.5-32B-Instruct | 37 | #339 |
| 15 | Qwen2.5-32B-Instruct-CFT | 37 | #375 |
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-ht-abstract-algebra · How It Works · Data refreshed daily, snapshot 2026-10-20.