Search pages, resources, and actions
Models, datasets, and services for materials science research.
Multi-head universal foundation interatomic potential covering inorganic crystals, molecules, surfaces, and reactive chemistry in a single model. Trained on OMAT-24 (100M PBE/PBE+U configurations) and fine-tuned across seven heads (omat_pbe, omol, oc20, spice, rgd1, mptrj, matpes). Predicts energies, forces, and stresses.
Foundation model for atomistic simulations. Universal machine learning interatomic potential trained on 150K Materials Project structures. Predicts energies, forces, and stresses for any material.
Universal potential for charge-informed atomistic modeling. Predicts energy, forces, stress, and magnetic moments.
