Open Arabic LLM - Arabic MMLU Biology (High School): leaderboard
The "Arabic MMLU Biology (High School)" 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 July 2028. 163 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | gemma-3-27B-pt | 58.98 | #502 |
| 2 | Qwen 2.5 72B Instruct | 58.62 | #454 |
| 3 | Llama 3.3 70B Instruct | 57.91 | #534 |
| 4 | Qwen2.5-32B-Instruct-abliterated-v2 | 55.86 | #394 |
| 5 | Hala-9B | 55.78 | #504 |
| 6 | AIC-1 | 55.64 | #231 |
| 7 | Qwen 2 72B | 55.29 | #521 |
| 8 | Qwen 2.5 32B Instruct | 54.29 | #498 |
| 9 | Saka-14B | 53.94 | #344 |
| 10 | Qwen2.5-32B-Instruct-CFT | 53.73 | #375 |
| 11 | Qwentile2.5-32B-Instruct | 53.3 | #339 |
| 12 | Awqward2.5-32B-Instruct | 53.23 | #371 |
| 13 | gemma-3-12B-pt | 52.66 | #686 |
| 14 | free-evo-qwen72B-v0.8-re | 52.31 | #435 |
| 15 | Fanar-1-9B-Instruct | 52.24 | #633 |
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-biology-high-school · How It Works · Data refreshed daily, snapshot 2026-10-20.