Open Arabic LLM - Arabic MMLU Geography (High School): leaderboard
The "Arabic MMLU Geography (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 January 2027. 163 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 2.5 72B Instruct | 66.86 | #454 |
| 2 | gemma-3-27B-pt | 62.91 | #502 |
| 3 | free-evo-qwen72B-v0.8-re | 62.04 | #435 |
| 4 | AceGPT-v2-70B-Chat | 61.56 | #407 |
| 5 | Llama 3.3 70B Instruct | 61.18 | #534 |
| 6 | AIC-1 | 60.98 | #231 |
| 7 | Saka-14B | 60.89 | #344 |
| 8 | Qwen2.5-32B-Instruct-abliterated-v2 | 60.6 | #394 |
| 9 | Qwentile2.5-32B-Instruct | 60.21 | #339 |
| 10 | Qwen 2.5 72B | 59.44 | #499 |
| 11 | AceGPT-v2-32B-Chat | 59.44 | #507 |
| 12 | tempmotacilla-cinerea-0308 | 58.86 | #395 |
| 13 | Qwen 2.5 32B Instruct | 58.77 | #498 |
| 14 | Qwen2.5-32B-Instruct-CFT | 58.48 | #375 |
| 15 | Awqward2.5-32B-Instruct | 58.38 | #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-geography-high-school · How It Works · Data refreshed daily, snapshot 2026-10-20.