Due to late diagnosis and the inability to find precancerous stages, lung cancer remains the leading cause of death from cancer. The research work presented in the article describes the application of vision transformers (ViTs) in generating an accessible and efficient method of early lung cancer screening. It accurately defines lung disorders as normal, benign, or malignant by the global attention ability of ViTs. Techniques such as data augmentation for better generalization, transfer learning to get maximum performance on medical datasets, and weighted loss functions in class imbalance have been used. We identified that the performance of the hybrid model of vision transformer with convolutional neural networks (CNNs) in the early detection of lung cancer outperforms other deep learning networks like standalone EfficientNet and convolutional neural networks, achieving a diagnostic accuracy of 94.1%. AI-assisted diagnosis realizes enhanced transparency and self-assurance through visual enlightenments with Grad-CAM as interpretable.