Open Arabic LLM - Alghafa Meta AR MSA: leaderboard
The "Alghafa Meta AR MSA" 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 | 91.73 | #231 |
| 2 | Qwen 2.5 72B Instruct | 91.51 | #454 |
| 3 | Qwentile2.5-32B-Instruct | 91.17 | #339 |
| 4 | Awqward2.5-32B-Instruct | 91.06 | #371 |
| 5 | Qwen 2.5 32B Instruct | 90.95 | #498 |
| 6 | Qwen2.5-32B-Instruct-CFT | 90.95 | #375 |
| 7 | Qwen2.5-32B-Instruct-abliterated-v2 | 90.84 | #394 |
| 8 | AceGPT-v2-70B-Chat | 90.5 | #407 |
| 9 | Hala-9B | 90.17 | #504 |
| 10 | Qwen 2.5 32B | 89.83 | #546 |
| 11 | Qwen 2 72B | 88.94 | #521 |
| 12 | Gemma 3 27B (IT) | 88.83 | #523 |
| 13 | AceGPT-v2-32B-Chat | 88.6 | #507 |
| 14 | Fanar-1-9B-Instruct | 88.49 | #633 |
| 15 | EZO-Qwen2.5-32B-Instruct | 88.27 | #471 |
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-meta-ar-msa · How It Works · Data refreshed daily, snapshot 2026-10-20.