Open Arabic LLM - Alghafa Average: leaderboard
The "Alghafa Average" 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 | AIC-1 | 79 | #231 |
| 2 | Qwen 2 72B | 78.56 | #521 |
| 3 | Qwen 2.5 72B Instruct | 78.22 | #454 |
| 4 | Hala-9B | 78.17 | #504 |
| 5 | Qwen 2.5 32B Instruct | 78.11 | #498 |
| 6 | Awqward2.5-32B-Instruct | 78.01 | #371 |
| 7 | Qwentile2.5-32B-Instruct | 78 | #339 |
| 8 | Qwen2.5-32B-Instruct-abliterated-v2 | 77.98 | #394 |
| 9 | Qwen2.5-32B-Instruct-CFT | 77.95 | #375 |
| 10 | Gemma 3 27B (IT) | 77.9 | #523 |
| 11 | aya-expanse-32B | 77.61 | #692 |
| 12 | AceGPT-v2-32B-Chat | 77.33 | #507 |
| 13 | Qwen 2.5 72B | 77.16 | #499 |
| 14 | Qwen 2.5 14B | 77.15 | #725 |
| 15 | Qwen 2.5 32B | 77.05 | #546 |
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-alghafa-average · How It Works · Data refreshed daily, snapshot 2026-10-20.