Open Arabic LLM - Arabic MMLU Arabic Language (Grammar): leaderboard
The "Arabic MMLU Arabic Language (Grammar)" 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: Around March 2027. 163 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | AIC-1 | 72.6 | #231 |
| 2 | Saka-14B | 71.23 | #344 |
| 3 | Fanar-1-9B-Instruct | 69.04 | #633 |
| 4 | AceGPT-v2-32B-Chat | 68.22 | #507 |
| 5 | free-evo-qwen72B-v0.8-re | 67.12 | #435 |
| 6 | Hala-9B | 66.85 | #504 |
| 7 | Llama 3.3 70B Instruct | 66.58 | #534 |
| 8 | Qwentile2.5-32B-Instruct | 61.37 | #339 |
| 9 | AceGPT-v2-70B-Chat | 60.27 | #407 |
| 10 | Qwen2.5-7B-Instruct-Uncensored | 54.25 | #684 |
| 11 | AceGPT-v2-32B | 53.97 | #768 |
| 12 | c4ai-command-r7B-arabic-02-2025 | 53.42 | #732 |
| 13 | SauerkrautLM-Nemo-12B-Instruct | 53.15 | #796 |
| 14 | aya-23-35B | 51.23 | #883 |
| 15 | tempmotacilla-cinerea-0308 | 51.23 | #395 |
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-arabic-language-grammar · How It Works · Data refreshed daily, snapshot 2026-10-20.