Open Arabic LLM - Arabic MMLU HT Machine Learning: leaderboard
The "Arabic MMLU HT Machine Learning" 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 2029. 163 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 2.5 72B Instruct | 56.25 | #454 |
| 2 | Qwentile2.5-32B-Instruct | 56.25 | #339 |
| 3 | Qwen 2.5 72B | 53.57 | #499 |
| 4 | Qwen 2 72B | 53.57 | #521 |
| 5 | AIC-1 | 53.57 | #231 |
| 6 | Awqward2.5-32B-Instruct | 52.68 | #371 |
| 7 | Llama 3.3 70B Instruct | 50.89 | #534 |
| 8 | Qwen2.5-32B-Instruct-CFT | 50.89 | #375 |
| 9 | Qwen 2.5 32B Instruct | 50 | #498 |
| 10 | Qwen 2.5 32B | 49.11 | #546 |
| 11 | Rombos-LLM-V2.6-Qwen-14B | 49.11 | #516 |
| 12 | Qwen3-8B-Base | 48.21 | #626 |
| 13 | Qwen2.5-32B-Instruct-abliterated-v2 | 48.21 | #394 |
| 14 | Fanar-1-9B | 47.32 | #778 |
| 15 | Gemma 3 27B (IT) | 46.43 | #523 |
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-machine-learning · How It Works · Data refreshed daily, snapshot 2026-10-20.