Open Arabic LLM - Arabic MMLU HT Sociology: leaderboard
The "Arabic MMLU HT Sociology" 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 | Llama 3.3 70B Instruct | 82.09 | #534 |
| 2 | Qwen 2.5 72B Instruct | 81.59 | #454 |
| 3 | AceGPT-v2-70B-Chat | 81.09 | #407 |
| 4 | gemma-3-27B-pt | 80.6 | #502 |
| 5 | Saka-14B | 79.6 | #344 |
| 6 | Gemma 2 27B (IT) | 79.1 | #645 |
| 7 | Qwen 2 72B | 78.61 | #521 |
| 8 | free-evo-qwen72B-v0.8-re | 78.61 | #435 |
| 9 | Gemma 3 27B (IT) | 78.11 | #523 |
| 10 | gemma-3-12B-pt | 78.11 | #686 |
| 11 | Fanar-1-9B | 77.61 | #778 |
| 12 | aya-expanse-32B | 77.11 | #692 |
| 13 | AceGPT-v2-32B-Chat | 76.12 | #507 |
| 14 | tempmotacilla-cinerea-0308 | 76.12 | #395 |
| 15 | Qwen 2.5 14B | 75.62 | #725 |
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-sociology · How It Works · Data refreshed daily, snapshot 2026-10-20.