Open Arabic LLM - Arabic MMLU HT Human Aging: leaderboard
The "Arabic MMLU HT Human Aging" 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 | AceGPT-v2-70B-Chat | 71.3 | #407 |
| 2 | Qwen 2.5 72B Instruct | 70.4 | #454 |
| 3 | free-evo-qwen72B-v0.8-re | 70.4 | #435 |
| 4 | Qwen 2.5 72B | 69.51 | #499 |
| 5 | Qwentile2.5-32B-Instruct | 69.51 | #339 |
| 6 | Gemma 3 27B (IT) | 69.06 | #523 |
| 7 | Qwen 2 72B | 69.06 | #521 |
| 8 | Llama 3.3 70B Instruct | 68.61 | #534 |
| 9 | gemma-3-27B-pt | 67.26 | #502 |
| 10 | Qwen2.5-32B-Instruct-abliterated-v2 | 67.26 | #394 |
| 11 | AceGPT-v2-32B-Chat | 66.82 | #507 |
| 12 | recoilme-gemma-2-9B-v0.4 | 66.37 | #553 |
| 13 | jais-family-30B-16k-chat | 65.92 | #755 |
| 14 | Hala-9B | 65.92 | #504 |
| 15 | Awqward2.5-32B-Instruct | 65.47 | #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-human-aging · How It Works · Data refreshed daily, snapshot 2026-10-20.