Mangoes are among the world's most sought-after fruits; therefore, they are very sensitive to many diseases that affect their production. Regular manual testing for diseases in the leaves of the mangoes is tedious, ineffective, and prone to errors. In this essay, I present the Hybrid InceptionV3 with Vision Transformer (ViT) architecture that can be used for disease diagnosis on the leaf of a mango tree. InceptionV3 can extract features in the images while the Vision Transformer would allow the network to grasp the whole context of the leaf. InceptionV3 can extract features in images while the Vision Transformer would allow the network to grasp the whole context of the leaf. By using LIME and Grad-CAM, interpretability of the AI model can be enhanced. By using LIME and Grad-CAM, interpretability of the AI model can be enhanced. Evaluations are performed on a dataset that consists of 3,000 images of six categories, and the model achieves an accuracy of 99.11%. It was evident that the proposed architecture performs better than other deep models in detecting diseases in the leaf of a mango tree.