Which Models Perform Better in Inheritance Reasoning?
A new paper evaluates the performance of commercial and open-source large language models on Arabic Islamic inheritance reasoning tasks. The study found that commercial models generally outperform open-source models, showing greater reliability in identifying heirs, applying exclusion rules, and maintaining consistency. Gemini 2.5 Flash achieved the best performance among the evaluated models, with a Mean Reciprocal Error (MRE) of 0.989. AI
IMPACT Highlights the current limitations of open-source models in complex legal and numerical reasoning, suggesting areas for future development.