One of the biggest challenges in the medical industry is accurately classifying brain illnesses using MRI images, especially when it comes to identifying particular tumor types such pituitary tumors, meningiomas, and gliomas. This paper introduces a sophisticated deep learning approach involving convolutional neural networks (CNN) to automatically and effectively classify brain cancers. The goal of the project is to improve diagnostic capabilities by determining the type of tumor and where it is located in the brain, in addition to forecasting whether a tumor will be present. The findings show that these models have the potential to greatly increase diagnostic accuracy for classifying brain tumors, providing physicians with a trustworthy instrument to accurately identify tumor kinds and the areas they affect. This strategy could significantly advance medical imaging and AI-driven healthcare solutions by improving early diagnosis and individualized therapy planning. The three main stages of the suggested methodology are performance evaluation, model training, and data preprocessing. To manage the intricacy of MRI data, we use data augmentation techniques and guarantee class balance at the preprocessing stage. We then use four deep learning models, each of which is used for its feature extraction and classification capabilities: CNN, MLP, Inception V3, and AlexNet. To guarantee accurate model evaluation over several datasets, let’s take a look how k-fold cross validation is done. Finally, we perform a comparative analysis amongst the models according to classification metrics including accuracy, precision, recall, F1 score, and AUC score.