IslamicTurathBench Evaluates LLMs on 12 Centuries of Islamic Scholarship
August 6, 2026
IslamicTurathBench introduces 3,465 question-answer items across 35 source works to assess model performance in classical Islamic studies. The benchmark uses a multi-task structure covering seven fields, categorized by scholarly demand levels from beginner to advanced.
HOW THIS AFFECTS YOU
●
builderYou can use this benchmark to evaluate RAG systems or fine-tuned models specialized in religious or historical domains.
●
researcherThis provides a high-quality, expert-reviewed dataset for studying domain-specific reasoning in LLMs.