Alzheimer is a hopeless neurodegenerative disease that mostly affects the memory of the brain of the elderly. All over the world, Alzheimer has an impact on people aged 65 and over. Early diagnosis leads to accurate detection of this disease. Due to the huge number of patients with this disease, manual diagnosis is error-prone and time-consuming. Different techniques have been used to classify Alzheimer’s disease, but more precision is needed to improve early detection, diagnosis and treatment solutions. In this paper, we proposed a deep learning based-model that deploys a new configuration of CNN (Convolutional Neural Networks) architecture for Alzheimer’s disease classification using brain MRI (Magnetic Resonance Imaging) scan. The proposed model classifies Alzheimer’s disease into non-dementia (ND), very mild dementia (VMD), mild dementia (MDTD) and moderate dementia (MD). Experiments show that the new CNN architecture is effective in the training and testing stages with promising results (recall = 91.7%). The proposed model can be used for real-time analysis and classification of Alzheimer’s disease.