Common lung diseases include pneumonia, pulmonary nodules, chronic obstructive pulmonary disease, and lung cancer. In recent years, with the development of artificial intelligence, large models have become increasingly prominent in the diagnosis of lung diseases. This paper employs a literature review approach to systematically the application results of large models in the diagnosis of lung nodules and lung cancer, discuss their technical advantages and value, and analyze the core problems and existing challenges. Studies have shown that large models exhibit unique value in pulmonary nodule and lung cancer diagnosis. Their core advantage is integrating images, text and clinical data. This optimizes the traditional computer-aided diagnosis (CAD) process. Specifically, integration with CAD systems improves the accuracy of benign-malignant differentiation of pulmonary nodules. Multimodal models effectively solve the problems of low contrast and blurred boundaries in PET/CT images for lung cancer diagnosis. Automatic report generation technology further helps improve clinical efficiency. However, large models still face many challenges in application, including clinical decision-making risks caused by the lack of metacognitive capabilities, the spread of misinformation caused by data poisoning, ethical issues such as training data bias and privacy leakage, and common problems such as unstable complex reasoning capabilities and disconnection between laboratory performance and clinical utility. Therefore, future research on large models in pulmonary disease diagnosis needs further deepening and expansion.