Open Arabic LLM - Arabic MMLU HT Anatomy: leaderboard
The "Arabic MMLU HT Anatomy" 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 January 2027. 163 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 2.5 72B Instruct | 60 | #454 |
| 2 | AceGPT-v2-70B-Chat | 59.26 | #407 |
| 3 | gemma-3-27B-pt | 58.52 | #502 |
| 4 | Gemma 3 27B (IT) | 55.56 | #523 |
| 5 | Llama 3.3 70B Instruct | 55.56 | #534 |
| 6 | AIC-1 | 54.81 | #231 |
| 7 | Qwentile2.5-32B-Instruct | 54.07 | #339 |
| 8 | Gemma 2 27B (IT) | 53.33 | #645 |
| 9 | Qwen 2.5 32B Instruct | 53.33 | #498 |
| 10 | Awqward2.5-32B-Instruct | 53.33 | #371 |
| 11 | Qwen2.5-32B-Instruct-CFT | 53.33 | #375 |
| 12 | AceGPT-v2-32B-Chat | 52.59 | #507 |
| 13 | Fanar-1-9B-Instruct | 51.85 | #633 |
| 14 | Qwen2.5-32B-Instruct-abliterated-v2 | 51.85 | #394 |
| 15 | aya-expanse-32B | 51.11 | #692 |
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-ht-anatomy · How It Works · Data refreshed daily, snapshot 2026-10-20.