arXiv:2605.19723v3 Announce Type: replace 
Abstract: Mathematical reasoning is essential for problem-solving in education, science, and industry, serving as a crucial benchmark for evaluating artificial intelligence systems. As Large Language Models (LLMs) improve their reasoning capabilities, understanding how well they perform mathematical reasoning has become increasingly important. This survey synthesizes recent advancements in mathematical reasoning with LLMs through a structured analysis of datasets, architectures, training strategies, and evaluation protocols. Our systematic review encompasses approximately 120 peer-reviewed studies and preprints, examining the evolution of this research area and providing a unified analytical framework to understand current progress and limitations. Our study particularly introduces a unified taxonomy of mathematical datasets, distinguishing between pretraining corpora, supervised fine-tuning resources, and evaluation benchmarks across varying levels of reasoning complexity. A systematic analysis of reasoning architectures and training strategies, including tool integration, verifier-guided reasoning, and parameter-efficient adaptation, is presented to assess their effects on reasoning robustness and generalization. Moreover, a comparative evaluation of existing metrics highlights the gap between final-answer accuracy and process-level reasoning verification. By synthesizing insights across these areas, our analysis identifies recurring failure modes, such as reasoning faithfulness issues, benchmark biases, and generalization limitations, and outlines key research directions toward improving symbolic grounding, evaluation reliability, and the development of more robust and trustworthy LLM-based reasoning systems.

تم نشر مسح شامل حول التفكير الرياضي في نماذج اللغة الكبيرة (LLMs)، حيث تم تجميع التقدم في مجموعات البيانات والهياكل واستراتيجيات التدريب وبروتوكولات التقييم. تغطي هذه المراجعة حوالي 120 دراسة تمت مراجعتها من قبل الأقران، مما يسلط الضوء على تطور هذا المجال والقيود الحالية.

Se ha publicado una encuesta integral sobre el razonamiento matemático en Modelos de Lenguaje de Gran Escala (LLMs), sintetizando avances en conjuntos de datos, arquitecturas, estrategias de entrenamiento y protocolos de evaluación. Esta revisión abarca alrededor de 120 estudios revisados por pares, destacando la evolución y las limitaciones actuales en el campo.

Une enquête complète sur le raisonnement mathématique dans les grands modèles de langage (LLMs) a été publiée, synthétisant les avancées dans les ensembles de données, les architectures, les stratégies d'entraînement et les protocoles d'évaluation. Cette revue couvre environ 120 études évaluées par des pairs, mettant en lumière l'évolution et les limitations actuelles dans le domaine.

A comprehensive survey on mathematical reasoning in Large Language Models (LLMs) has been published, synthesizing advancements in datasets, architectures, training strategies, and evaluation protocols. This review encompasses around 120 peer-reviewed studies, highlighting the evolution and current limitations in the field.