Open Arabic LLM - ALL — leaderboard
Metric: Accuracy (%). Source: huggingface.co. 163 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Rombos-LLM-V2.5-Qwen-72B | 75.58 |
| 2 | Ultiima-72B | 75.57 |
| 3 | calme-2.1-qwen2.5-72B | 73.44 |
| 4 | calme-2.2-qwen2.5-72B | 73.26 |
| 5 | Qwen 2.5 72B Instruct | 73.18 |
| 6 | gemma-3-27B-pt | 69.96 |
| 7 | lambda-qwen2.5-32B-dpo-test | 69.81 |
| 8 | Qwen 2 72B | 69.48 |
| 9 | Qwentile2.5-32B-Instruct | 69.45 |
| 10 | Llama 3.3 70B Instruct | 69.32 |
| 11 | Qwen2.5-32B-Instruct-abliterated-v2 | 69.2 |
| 12 | Saka-14B | 69.18 |
| 13 | Awqward2.5-32B-Instruct | 69.02 |
| 14 | Gemma 3 27B (IT) | 68.7 |
| 15 | Qwen2.5-32B-Instruct-CFT | 68.62 |
Interactive version: theaggregate.ai/benchmark?slug=open-arabic-llm-all · How the rankings work · Data refreshed daily, snapshot 2026-07-22.