Open Arabic LLM - Arabic MMLU HT Moral Scenarios: leaderboard
The "Arabic MMLU HT Moral Scenarios" 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 2028. 163 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Llama 3.3 70B Instruct | 64.36 | #534 |
| 2 | EZO-Qwen2.5-32B-Instruct | 58.55 | #471 |
| 3 | Qwen 2.5 32B Instruct | 58.1 | #498 |
| 4 | Qwen2.5-32B-Instruct-CFT | 57.54 | #375 |
| 5 | Awqward2.5-32B-Instruct | 56.98 | #371 |
| 6 | Qwen 2.5 72B Instruct | 55.08 | #454 |
| 7 | AceGPT-v2-70B-Chat | 52.63 | #407 |
| 8 | Saka-14B | 51.96 | #344 |
| 9 | Rombos-LLM-V2.6-Qwen-14B | 51.62 | #516 |
| 10 | Qwen 2 72B | 51.06 | #521 |
| 11 | Qwen2.5-32B-Instruct-abliterated-v2 | 50.61 | #394 |
| 12 | lambda-qwen2.5-14B-dpo-test | 50.06 | #481 |
| 13 | tempmotacilla-cinerea-0308 | 49.61 | #395 |
| 14 | Qwentile2.5-32B-Instruct | 48.38 | #339 |
| 15 | AIC-1 | 47.6 | #231 |
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-moral-scenarios · How It Works · Data refreshed daily, snapshot 2026-10-20.