The survey explores advancements in large language models (LLMs) focusing on their reasoning capabilities. It discusses the architecture improvements, training methodologies, and the challenges of incorporating reasoning into LLMs. Key points include the significance of multi-modal inputs, contextual understanding, and the need for robust evaluation metrics that truly assess reasoning ability. The survey also highlights open challenges like addressing biases, improving generalization across tasks, and the potential for LLMs to assist in solving complex problems. Overall, the research indicates a growing optimism about enhancing reasoning capabilities in LLMs, which could revolutionize AI applications in various domains such as education, healthcare, and law.