Open Arabic LLM - Arabic MMLU HT Econometrics: leaderboard
The "Arabic MMLU HT Econometrics" 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 December 2027. 163 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 2.5 72B Instruct | 57.02 | #454 |
| 2 | Llama 3.3 70B Instruct | 53.51 | #534 |
| 3 | Gemma 2 27B (IT) | 53.51 | #645 |
| 4 | Saka-14B | 53.51 | #344 |
| 5 | Qwen 2 72B | 51.75 | #521 |
| 6 | Gemma 3 27B (IT) | 50.88 | #523 |
| 7 | Qwen 2.5 72B | 50.88 | #499 |
| 8 | AIC-1 | 50.88 | #231 |
| 9 | AceGPT-v2-70B-Chat | 49.12 | #407 |
| 10 | Qwen3-8B-Base | 48.25 | #626 |
| 11 | Qwen2.5-32B-Instruct-abliterated-v2 | 47.37 | #394 |
| 12 | tempmotacilla-cinerea-0308 | 47.37 | #395 |
| 13 | Qwen 2.5 32B Instruct | 46.49 | #498 |
| 14 | Qwen 2.5 14B | 46.49 | #725 |
| 15 | Awqward2.5-32B-Instruct | 46.49 | #371 |
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-econometrics · How It Works · Data refreshed daily, snapshot 2026-10-20.