A new GenAI model from Latent Labs generates lab-ready protein binders in seconds, drastically cutting time and cost in drug discovery.
Latent Labs has launched Latent-X, a GenAI model designed for push-button protein binder design. Accessible via a no-code web platform, it allows scientists to upload protein targets and generate macrocycles and mini-binders instantly. Traditional screening processes take months and cost thousands, with hit rates below 1%. Latent-X reduces that to minutes and achieves 91–100% hit rates for macrocycles and 10–64% for mini-binders—a breakthrough for preclinical drug design.
Built to solve structural binding puzzles at the atomic level, Latent-X produces all-atom resolution outputs. The model co-samples structure and sequence simultaneously, enabling ultra-fast binder generation and testing. It also generalizes well to previously untargeted proteins, supporting novel drug discovery pathways beyond known biological repertoires. In head-to-head lab tests, it outperformed all previous generative models for protein binders.
The platform includes structure visualization tools, target hotspot selectors, and automated ranking metrics. Researchers can select top-ranked candidates for synthesis without needing AI or wet lab infrastructure. This system is over 10x faster than earlier workflows and enables early-stage R&D teams to design with precision and speed. Users can generate strong binders for diverse therapeutic targets at a fraction of the cost.
Latent Labs is positioning this tool to make biology programmable. The company recently raised $50M, and its team includes former AlphaFold 2 developers and AI leaders from DeepMind and Stability AI. Latent-X marks a major leap toward computational-first drug design, where drug discovery happens in silico before entering the lab.
