Brain tumors are a serious health issue that thousands of people worldwide are affected. For brain tumor to be treated successfully and patients have an enhance quality of life. Thus, early and precise tumor diagnosis is essential. Several imaging methods are employed for identification of brain tumors. The most popular of these methods is MRI scan. In order to get around the drawback of conventional methods, computer -aided analysis of brain picture has emerged as a viable method for precise and accurate brain tumor classification in recent years. In the current study, we constructed an improved vision transformer model that correctly detects and classifies brain tumors. This model was able to classify the brain tumors, when it was trained using MRI dataset, which included 5712 pictures of the tumors. The accuracy of Vision Transformer (ViT) model was 98.13%. To diagnose rarer and more complicated tumors with greater accuracy, further research may be done.