IslamicMMLU - Fiqh: leaderboard
Metric: Accuracy (%) on IslamicMMLU's 3,200 Fiqh knowledge questions from the four-school encyclopedia of al-Jaziri (the 800 madhab-bias questions are scored separately); four-option multiple choice in Arabic, zero-shot, temperature 0, answer letter only (10 new tokens), via each model's API; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 26 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Flash | 89.1 | #93 |
| 2 | Gemini 3 Pro | 87.9 | #77 |
| 3 | GPT-5.2 | 85.9 | #105 |
| 4 | Gemini 2.5 Pro | 85.5 | #145 |
| 5 | Claude Sonnet 4.5 | 84.9 | #138 |
| 6 | GPT-5 | 83.5 | #91 |
| 7 | Gemini 2.5 Flash | 83.1 | #237 |
| 8 | GPT-5.1 | 81.5 | #131 |
| 9 | Claude 3.7 Sonnet | 81.1 | #241 |
| 10 | GPT-4o | 78.5 | #333 |
| 11 | GPT-4.1 | 77.5 | #240 |
| 12 | DeepSeek V3.2 | 74.3 | #198 |
| 13 | Grok 4.1 Fast | 74 | #208 |
| 14 | Claude Opus 4.5 | 72.7 | #79 |
| 15 | Llama 4 Scout | 70.6 | #646 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=islamicmmlu-fiqh · How It Works · Data refreshed daily, snapshot 2026-10-11.