This paper provides a comprehensive survey of large language models (LLMs) in education (LLMEdu), discussing their current state, key technologies, challenges, and future development. LLMs, with their deep learning, pre-training, fine-tuning, and reinforcement learning capabilities, have become powerful tools for smart education. They are used in various applications such as natural language processing, computer vision, speech recognition, and autonomous driving. In education, LLMs can provide personalized, adaptive learning experiences, enhance immediate feedback, and support intelligent teaching. However, they still face challenges such as data privacy, security, and the need for further optimization. The paper discusses the characteristics of LLMs and education, the process of integrating LLMs into education, and the key technologies involved. It also highlights the problems and challenges in LLMEdu and proposes future directions for its development. The integration of LLMs into education has the potential to revolutionize the learning experience by providing personalized support, expanding knowledge coverage, promoting critical thinking, and enhancing the accessibility and inclusivity of education. The paper concludes that LLMs are one of the most important technologies in human intelligence and have the potential to support the sustainable development of education.This paper provides a comprehensive survey of large language models (LLMs) in education (LLMEdu), discussing their current state, key technologies, challenges, and future development. LLMs, with their deep learning, pre-training, fine-tuning, and reinforcement learning capabilities, have become powerful tools for smart education. They are used in various applications such as natural language processing, computer vision, speech recognition, and autonomous driving. In education, LLMs can provide personalized, adaptive learning experiences, enhance immediate feedback, and support intelligent teaching. However, they still face challenges such as data privacy, security, and the need for further optimization. The paper discusses the characteristics of LLMs and education, the process of integrating LLMs into education, and the key technologies involved. It also highlights the problems and challenges in LLMEdu and proposes future directions for its development. The integration of LLMs into education has the potential to revolutionize the learning experience by providing personalized support, expanding knowledge coverage, promoting critical thinking, and enhancing the accessibility and inclusivity of education. The paper concludes that LLMs are one of the most important technologies in human intelligence and have the potential to support the sustainable development of education.