DNA-MP is an Artificial Intelligence-based tool that makes use of a novel DNA sequence encoder and a deep forest classifier to more precisely and accurately predict 5-Hydroxymethylcytosine (5hmc), 4-Methylcytosine (4mc), and N6-methyladenine (6mA) modifications only using raw DNA sequences. DNA-MP significantly outperforms state-of-the-art 4mc, 5hmc, and 6mA predictors on 17 benchmark datasets related to 12 different species in k-fold cross-validation and independent test sets based evaluation. DNA-MP web interface enables the user to perform multi-dimensional analysis of DNA sequences, training and optimizing the machine learning model from scratch, using pre-trained models of different species to make inferences on new sequences, and download interactive artifacts during the lifetime of the session.