1910, the AI-native biotech company at the forefront of small and large molecule therapeutics discovery, has announced the publication of PEGASUS™, a groundbreaking multimodal AI model designed to predict and generate novel drug-like macrocyclic peptides. The research, featured in the Journal of Medicinal Chemistry, marks a significant advancement in peptide-based drug discovery by achieving state-of-the-art accuracy in designing macrocyclic peptides capable of crossing cell membranes. PEGASUS™ is the first AI model to successfully predict and generate macrocyclic peptides containing more than two polar or charged functional groups—features traditionally associated with poor cell permeability—while still demonstrating measurable in vitro cell penetration. This breakthrough addresses a long-standing challenge in drug development, where macrocyclic peptides offer high target specificity and potency but often fail to enter cells due to their polar nature. By integrating multimodal data—including molecular structure, physicochemical properties, and biological activity—PEGASUS™ leverages deep learning to navigate the complex chemical space of macrocyclic peptides. The model enables rapid, rational design of novel candidates with optimized drug-like properties, significantly accelerating the path from concept to preclinical candidate. 1910’s work with PEGASUS™ exemplifies its mission to apply AI not just to analyze data, but to actively engineer new therapeutics. The company’s AI-native platform is designed to discover and optimize both small and large molecules, with a particular focus on targets that have been historically difficult to drug. The publication of PEGASUS™ underscores 1910’s leadership in AI-driven drug discovery and opens new avenues for developing peptide-based therapies against challenging disease targets, including intracellular proteins and protein-protein interactions.