Broadening access to Skala creates a faster path to predictive DFT

Microsoft Research releases Skala 1.1, a deep-learning density functional theory functional trained on more data than its predecessor. It integrates into CP2K, Psi4, FHI-aims, ORCA, and VASP. The model ranks first in thirty-two of fifty-five GMTKN55 benchmark categories, outperforming expensive global hybrid functionals at meta-GGA computational cost.