As readers may know, many open-source molecular generative models have become available recently.A few years ago, SMILES-based generative models were dominant, but diffusion-based models have recently become mainstream. Today, I would like to introduce EvoDiffMol, a new conditional diffusion-based generative model recently published in the Journal of Cheminformatics. The model can generate molecules in 3D, conditioned on user-specified target properties, for example user-specified scaffold, ADMET properties such as CYP, hERG, Caco-2 etc.