AI Model Ruby Predicts Drug Potency Faster And More Accurately
A new artificial intelligence model called Ruby is changing how scientists predict whether a drug molecule will work. The model was described on November 20 by Genesis Molecular AI and it focuses on potency. In tests on four public benchmarks Ruby reached the highest average correlation of all methods checked including physics based free energy methods. It also performed best on eight real assays from five collaboration programs with Incyte. Ruby reuses structural knowledge from an earlier foundation model called Pearl instead of building a fresh three dimensional pose for every compound. That shortcut lets it score a compound in 25 milliseconds which is 40 to more than 1000 times faster per compound than other methods. The team is also releasing GenesisPotency 2026 a cleaned public dataset so other researchers can trace each measurement back to its source.