Open Arabic LLM - Arabic MMLU HT Marketing: leaderboard
The "Arabic MMLU HT Marketing" 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 | gemma-3-27B-pt | 88.03 | #502 |
| 2 | Gemma 3 27B (IT) | 86.32 | #523 |
| 3 | AceGPT-v2-70B-Chat | 86.32 | #407 |
| 4 | Qwen 2 72B | 85.04 | #521 |
| 5 | AIC-1 | 85.04 | #231 |
| 6 | Llama 3.3 70B Instruct | 83.76 | #534 |
| 7 | Qwen 2.5 72B Instruct | 83.33 | #454 |
| 8 | Qwen 2.5 72B | 83.33 | #499 |
| 9 | Qwentile2.5-32B-Instruct | 82.91 | #339 |
| 10 | Awqward2.5-32B-Instruct | 81.2 | #371 |
| 11 | free-evo-qwen72B-v0.8-re | 81.2 | #435 |
| 12 | Gemma 2 27B (IT) | 80.77 | #645 |
| 13 | Qwen 2.5 32B Instruct | 80.77 | #498 |
| 14 | gemma-3-12B-pt | 80.77 | #686 |
| 15 | AceGPT-v2-32B-Chat | 80.77 | #507 |
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-marketing · How It Works · Data refreshed daily, snapshot 2026-10-20.