Latent Labs has unveiled Latent-X2, a next-generation AI model capable of generating drug-like biologics—specifically antibodies and peptides—without the need for iterative lab testing. The platform enables scientists to design high-affinity binders for challenging disease targets across multiple formats including VHH, scFv, and macrocyclic peptides, with drug-like properties built into the initial design. Latent-X2 achieves promising results from the first generation of designs. It successfully generated hits against half of 18 diverse and difficult targets, achieving picomolar to nanomolar binding affinities, often requiring only 4 to 24 designs. Notably, the model has demonstrated the ability to generate macrocyclic peptides that bind K-Ras—a target long considered undruggable—matching or surpassing results from trillion-scale mRNA display screens while testing 11 orders of magnitude fewer sequences. A key advancement is the model’s ability to produce molecules with inherent developability and low immunogenicity. In the first comprehensive evaluation of AI-generated antibodies for immunogenicity, de novo VHH binders showed potent target engagement with minimal activation of human T-cells and low cytokine release across a panel of ten donors. These results suggest that AI-designed molecules can now meet critical preclinical safety and stability benchmarks from the outset, significantly reducing the need for time-consuming optimization. The Latent Labs Platform, powered by Latent-X2, offers a user-friendly, no-code interface accessible via web browser or through API integration, allowing researchers to generate and explore novel biologics for their target of interest. The platform is designed to accelerate the path from hit to clinical candidate by minimizing wet lab work and reducing the risk of late-stage failure. Simon Kohl, CEO and founder of Latent Labs, said the company is bringing the same paradigm shift seen in semiconductors, satellites, and aircraft—where complex systems are now designed computationally before physical fabrication—to drug discovery. “With Latent-X2, we’re moving toward designing the right molecule from the start,” he said. Latent Labs has also announced the addition of Stefan Oschmann, former CEO of Merck KGaA, to its strategic advisory board. Oschmann praised the company’s approach, noting that it challenges decades of traditional drug discovery by enabling design from first principles rather than iterative refinement. Access to Latent-X2 is currently limited to selected partners. Interested parties can express interest at [email protected]. Latent-X2 builds on the success of Latent-X1, released just five months prior, which has already been adopted by leading academic and industry groups. The company recently raised $50 million in a funding round co-led by Radical Ventures and Sofinnova Partners, with participation from prominent figures including Anthropic’s Dario Amodei, Eleven Labs’ Mati Staniszewski, and Google’s Jeff Dean. The team includes former AlphaFold 2 co-developers and ex-DeepMind leads with experience from Microsoft, Apple, Exscientia, Mammoth Bio, Altos Labs, and Zymergen. For commercial inquiries, contact [email protected]. Scientific questions can be directed to the company’s support team. The platform is designed with safety and responsible use in mind, and Latent Labs implements safeguards to ensure ethical application of its technology.