Open Arabic LLM - Arabic MMLU Geography (Middle School): leaderboard
The "Arabic MMLU Geography (Middle 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: Estimated already saturated. 163 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 2.5 14B | 75 | #725 |
| 2 | Saka-14B | 75 | #344 |
| 3 | gemma-3-27B-pt | 74.26 | #502 |
| 4 | Qwen2.5-32B-Instruct-abliterated-v2 | 74.26 | #394 |
| 5 | AceGPT-v2-32B-Chat | 73.9 | #507 |
| 6 | AIC-1 | 73.53 | #231 |
| 7 | Llama 3.3 70B Instruct | 72.79 | #534 |
| 8 | Qwen 2.5 72B Instruct | 72.79 | #454 |
| 9 | Awqward2.5-32B-Instruct | 72.79 | #371 |
| 10 | tempmotacilla-cinerea-0308 | 72.79 | #395 |
| 11 | Rombos-LLM-V2.6-Qwen-14B | 72.79 | #516 |
| 12 | Qwen 2.5 32B Instruct | 72.43 | #498 |
| 13 | EZO-Qwen2.5-32B-Instruct | 72.43 | #471 |
| 14 | Qwen2.5-32B-Instruct-CFT | 72.06 | #375 |
| 15 | Qwen 2 72B | 71.69 | #521 |
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-middle-school · How It Works · Data refreshed daily, snapshot 2026-10-20.