IslamicMMLU - Verse Retrieval: leaderboard
Metric: Accuracy (%) on IslamicMMLU's 1,066 Quran track questions on giving the text of a verse from its surah and verse number, with trap positions past the end of the surah; 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: Estimated already saturated. 26 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Flash | 99.1 | #93 |
| 2 | GPT-5 | 98.9 | #91 |
| 3 | GPT-5.2 | 98.4 | #105 |
| 4 | GPT-5.1 | 97.5 | #131 |
| 5 | Gemini 3 Pro | 96.9 | #77 |
| 6 | Gemini 2.5 Pro | 96.6 | #145 |
| 7 | Claude Opus 4.5 | 93 | #79 |
| 8 | Claude Sonnet 4.5 | 85.4 | #138 |
| 9 | Grok 4.1 Fast | 78.9 | #208 |
| 10 | Gemini 2.5 Flash | 78.1 | #237 |
| 11 | Claude 3.7 Sonnet | 77.3 | #241 |
| 12 | GPT-4o | 72.5 | #333 |
| 13 | GPT-4.1 | 67.3 | #240 |
| 14 | Llama 4 Maverick | 46.8 | #451 |
| 15 | Claude Haiku 4.5 | 44.6 | #271 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=islamicmmlu-verse-retrieval · How It Works · Data refreshed daily, snapshot 2026-10-11.