Open Arabic LLM - Arabic MMLU Driving Test: leaderboard
The "Arabic MMLU Driving Test" 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 | 77.21 | #454 |
| 2 | AIC-1 | 76.88 | #231 |
| 3 | Qwen 2 72B | 76.8 | #521 |
| 4 | gemma-3-27B-pt | 76.14 | #502 |
| 5 | Saka-14B | 75.89 | #344 |
| 6 | Gemma 3 27B (IT) | 75.47 | #523 |
| 7 | Qwentile2.5-32B-Instruct | 74.81 | #339 |
| 8 | AceGPT-v2-32B-Chat | 74.48 | #507 |
| 9 | AceGPT-v2-70B-Chat | 74.48 | #407 |
| 10 | tempmotacilla-cinerea-0308 | 73.82 | #395 |
| 11 | Qwen3-8B-Base | 73.33 | #626 |
| 12 | Qwen2.5-32B-Instruct-abliterated-v2 | 73.33 | #394 |
| 13 | free-evo-qwen72B-v0.8-re | 73.25 | #435 |
| 14 | Qwen 2.5 14B | 73 | #725 |
| 15 | Hala-9B | 72.5 | #504 |
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-arabic-mmlu-driving-test · How It Works · Data refreshed daily, snapshot 2026-10-20.