Research & Papers

Microsoft's Skala 1.1 boosts DFT accuracy 2.5x with broader access

⚑Skala 1.1 achieves 2.8 kcal/mol error on GMTKN55, beating hybrid functionals at semi-local costs...

Deep Dive

Key Points
  • Skala 1.1 achieves 2.8 kcal/mol error on GMTKN55, surpassing hybrid functionals at semi-local computational costs.
  • Now integrated into CP2K and coming to Psi4, FHI-aims, ORCA, and VASP, broadening accessibility for chemists and material scientists.
  • Launches a living benchmark to track Skala’s performance, fostering community-driven optimization and transparency.

Why It Matters

Predictive DFT simulations just got faster and more accessible, accelerating drug discovery, materials science, and industrial R&D workflows.

πŸ“¬ Get the top 10 AI stories daily