ToxinPred3.0 is an upgraded version of ToxinPred (See PLoS One. 2013 Sep 13;8(9):e73957) to predict toxicity of peptides. It is based on mechine learning and deep learning methods that has been trained on a large dataset of experimentally validated toxic and non-toxic peptides which contains 5518 toxic and 5518 non-toxic peptides. It utilizes wide range of information and techniques for prediction, including machine learning, deep learning and MERCI. To use ToxinPred3.0 user can access the prediction module and simply enter the peptide sequence into the text box or upload a FASTA file and click on submit button. The server will then return a prediction of whether your peptide is toxic or non-toxic. In addition to the prediction module, there are some other useful modues for users, such as Design, Protein scanning, Motif scan and BLAST search. Furthermore, the server provides following important links to the users, including options to Download the standalone version of ToxinPred3.0 and the associated datasets. The Algorithm link provides details about the underlying methodologies employed by ToxinPred3.0. The Help page offers assistance and guidance on using the tool effectively. The Team link provides information about the team behind ToxinPred3.0, and the Contact link allows users to reach out for any inquiries or feedback. Lastly, if users are interested in predicting the toxicity of proteins rather than peptides, they should use our old server, ToxinPred2 (See Brief Bioinform. doi: 10.1093/bib/bbac174), which is specifically designed for protein toxicity prediction.