The study presents CerevianNet, a lightweight proprietary convolutional neural network (CNN) for multi-class brain tumor classification that reduces the number of parameters and computational complexity while maintaining high accuracy. The model was tested on five different datasets where it performed best on Dataset-5 with 99.67% training accuracy, 98.17% validation accuracy and 98.30% testing accuracy. He showed the lowest testing accuracy on Dataset-3 with 75.63%. Comparison with pre-trained models showed that EfficientNetb3 has the highest accuracy of 99.11%, while our own lightweight CNN achieved 98% accuracy with 4.1 times fewer parameters and less training time. The model performed well on larger data sets, but less successfully on smaller and unbalanced ones. The framework is optimized for small format devices and suitable for integration into clinical procedures for rapid diagnosis of brain tumors.