Open Arabic LLM - Alghafa Meta AR MSA — leaderboard
Metric: Accuracy (%). Source: huggingface.co. 163 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | lambda-qwen2.5-32B-dpo-test | 92.07 |
| 2 | Qwen 2.5 72B Instruct | 91.51 |
| 3 | Rombos-LLM-V2.5-Qwen-72B | 91.51 |
| 4 | calme-2.1-qwen2.5-72B | 91.4 |
| 5 | calme-2.2-qwen2.5-72B | 91.4 |
| 6 | Ultiima-72B | 91.4 |
| 7 | Qwentile2.5-32B-Instruct | 91.17 |
| 8 | Awqward2.5-32B-Instruct | 91.06 |
| 9 | Qwen 2.5 32B Instruct | 90.95 |
| 10 | Qwen2.5-32B-Instruct-CFT | 90.95 |
| 11 | Qwen2.5-32B-Instruct-abliterated-v2 | 90.84 |
| 12 | Qwen 2.5 32B | 89.83 |
| 13 | Qwen 2 72B | 88.94 |
| 14 | Gemma 3 27B (IT) | 88.83 |
| 15 | EZO-Qwen2.5-32B-Instruct | 88.27 |
Interactive version: theaggregate.ai/benchmark?slug=open-arabic-llm-alghafa-meta-ar-msa · How the rankings work · Data refreshed daily, snapshot 2026-07-22.