MoleculeMind announces QuantaMind, an AI-driven reactive atomistic modeling platform, with a Science Advances publication showing DFT-level accuracy in tens-of-nanosecond reactive MD simulations of systems containing up to 10,000 atoms.

QuantaMind is a transition-state-centered machine learning force field that reliably models bond making and breaking during long simulations, incorporating non-equilibrium conformations and transition-state data to learn key reaction pathway states.

The platform enables simulating an entire enzyme reaction at atomic scale, including proton transfer and full catalytic cycles, and has been scaled to reactive systems with hundreds of thousands of atoms for industrial applications.

Beyond predictions, QuantaMind integrates AI-driven generation with physics-based simulation to deliver explainable, verifiable mechanistic insights for drug discovery and enzyme engineering.

The framework uses DFT method classification embeddings to unify quantum chemistry data across functionals and basis sets, with benchmarks showing a 100,000-atom reactive system time-step around 0.25 seconds, signaling state-of-the-art performance among leading MLFFs.