QIMMA - ArabCulture: leaderboard
ArabCulture multiple-choice commonsense questions about everyday life and culture across Arab countries, as run in the QIMMA Arabic leaderboard. Scored with LightEval log-likelihood over the answer options with character-length normalization rather than generated answers, so reasoning-tuned models can land near chance; accuracy 0-100, higher is better.
Metric: Accuracy (%, log-likelihood multiple choice). Source: huggingface.co. Saturation forecast: Not forecast. 83 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Jais-2-70B-Chat | 83.24 | #430 |
| 2 | AceGPT-v2-32B-Chat | 79.79 | #518 |
| 3 | Llama 3.3 70B Instruct | 78.05 | #541 |
| 4 | jais-adapted-70B-chat | 78 | #639 |
| 5 | AIC-1 | 77.52 | #235 |
| 6 | Qwen 2.5 32B Instruct | 75.8 | #509 |
| 7 | Yehia-7B-preview | 75.47 | #659 |
| 8 | Hala-9B | 72.83 | #517 |
| 9 | jais-adapted-13B-chat | 72.16 | #948 |
| 10 | SILMA-9B-Instruct-v1.0 | 71.22 | #927 |
| 11 | Fanar-1-9B-Instruct | 70.71 | #635 |
| 12 | granite-4.0-h-3B-ar | 69.97 | #1002 |
| 13 | Jais-2-8B-Chat | 68.99 | #802 |
| 14 | LFM2 24B A2B | 68.83 | #1362 |
| 15 | c4ai-command-r7B-arabic-02-2025 | 66.94 | #727 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=qimma-arabculture · How It Works · Data refreshed daily, snapshot 2026-11-05.