Open Arabic LLM - Arabic MMLU HT Philosophy — leaderboard
Metric: Accuracy (%). Source: huggingface.co. 163 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | calme-2.1-qwen2.5-72B | 76.85 |
| 2 | calme-2.2-qwen2.5-72B | 76.53 |
| 3 | Rombos-LLM-V2.5-Qwen-72B | 76.21 |
| 4 | Ultiima-72B | 76.21 |
| 5 | Qwen 2.5 72B Instruct | 74.28 |
| 6 | gemma-3-27B-pt | 71.06 |
| 7 | Qwen2.5-32B-Instruct-abliterated-v2 | 70.74 |
| 8 | Llama 3.3 70B Instruct | 69.13 |
| 9 | Awqward2.5-32B-Instruct | 68.81 |
| 10 | lambda-qwen2.5-32B-dpo-test | 68.81 |
| 11 | gemma-3-12B-pt | 68.49 |
| 12 | Qwen 2.5 32B Instruct | 67.52 |
| 13 | Qwen2.5-32B-Instruct-CFT | 67.52 |
| 14 | Qwen 2 72B | 67.2 |
| 15 | Gemma 3 27B (IT) | 66.88 |
Interactive version: theaggregate.ai/benchmark?slug=open-arabic-llm-arabic-mmlu-ht-philosophy · How the rankings work · Data refreshed daily, snapshot 2026-07-22.