Open Arabic LLM - Madinah QA Arabic Language (General): leaderboard
The "Madinah QA Arabic Language (General)" 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 | Qwen 2.5 72B Instruct | 82.52 | #454 |
| 2 | gemma-3-27B-pt | 81.21 | #502 |
| 3 | AIC-1 | 81.21 | #231 |
| 4 | Qwen 2.5 32B Instruct | 81.05 | #498 |
| 5 | Awqward2.5-32B-Instruct | 81.05 | #371 |
| 6 | Qwen2.5-32B-Instruct-CFT | 80.56 | #375 |
| 7 | Gemma 3 27B (IT) | 80.23 | #523 |
| 8 | tempmotacilla-cinerea-0308 | 80.07 | #395 |
| 9 | lambda-qwen2.5-14B-dpo-test | 79.58 | #481 |
| 10 | Qwen2.5-32B-Instruct-abliterated-v2 | 79.41 | #394 |
| 11 | Qwentile2.5-32B-Instruct | 79.25 | #339 |
| 12 | Qwen 2.5 72B | 79.08 | #499 |
| 13 | EZO-Qwen2.5-32B-Instruct | 78.92 | #471 |
| 14 | Saka-14B | 78.92 | #344 |
| 15 | Qwen 2.5 14B | 78.43 | #725 |
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-madinah-qa-arabic-language-general · How It Works · Data refreshed daily, snapshot 2026-10-20.