By Nicole Sonnert
Today, Infinimmune released a preprint on bioRxiv outlining a generative protein language model that is core to their novel, human-first approach to antibody engineering. Infinimmune is built on the premise that there is no better starting point than humans to engineer therapeutic antibodies. The team has harnessed the vast inter-individual diversity of human antibody repertoires to develop a novel generative protein language model, trained on full-length, paired heavy and light chain antibody sequencing data derived from donated human blood.
Infinimmune’s AI model GLIMPSE-1, announced today, transforms how therapeutic antibodies are engineered. Most commercial antibodies come from model systems like in vivo immunization or display technologies. These approaches miss the sophisticated selection pressures that shape real immune responses. GLIMPSE-1 bridges this gap by learning from millions of human antibodies that have passed nature’s own quality control. GLIMPSE-1 can take antibodies from any source and enhance them with the collective wisdom of human immunity, incorporating patterns from autoantigen recognition that balance exquisite specificity with self-tolerance. The ability to generalize these uniquely human features to any antibody lets Infinimmune start with the best antibodies nature has evolved, regardless of source, and transform them into optimized human therapeutics. GLIMPSE-1 is already powering Infinimmune’s therapeutic pipeline, with their lead program in skin autoimmunity entering first-in-human trials in 2026.
Key results from the preprint include:
- Best-in-class performance in antibody humanization
- Successful affinity optimization achieving up to 1000-fold improvements across multiple targets
- Generation of sequence-divergent functional antibody variants
- Engineering for species cross-reactivity, including simultaneous optimization for human and cynomolgus monkey binding
- Elimination of developability liabilities while engineering for formulation-preferred properties like isoelectric point (pI)
Monoclonal antibody therapies specifically target disease-mediating proteins and are transforming patient care across a wide range of indications including cancer, autoimmune, and cardiometabolic diseases. As a result, patients are seeing longer survival, better disease control, and improved quality of life. However, traditional methods for engineering antibodies rely on relatively simplistic models that cannot account for the complexity of antibody sequence selection in the context of the human immune system.
The release of GLIMPSE-1 coincides with evolving FDA guidance that promotes human-relevant and non-animal approaches for monoclonal antibody development. Infinimmune’s platform produces human-optimized therapeutics from day one–positioning the company at the forefront of modern regulatory thinking.
As Dr. Wyatt Mc Donnell, co-founder and C E O of Infinimmune, states in today’s announcement, “Fully human antibodies carry the evolutionary logic of the immune system—optimized over millions of years. With GLIMPSE-1, we can decode that logic directly from immune repertoires to design better, safer biologics from day one. As the F D A shifts away from animal testing, models like GLIMPSE-1 will be critical to discovering and developing next-generation antibody therapies.”
GLIMPSE-1 has already demonstrated its real-world impact by engineering the antibodies in Infinimmune’s therapeutic pipeline. Infinimmune’s approach represents a fundamental shift in how therapeutic antibodies are created. By learning exclusively from the vast diversity of human immune repertoires, GLIMPSE-1 doesn’t just humanize antibodies. It designs them to harness the power of human immunology from the start.
The company plans to make the GLIMPSE-1 generative protein language model available to select academic and industry partners, while continuing to advance its internal pipeline of Complete Human® antibody therapeutics.
Nicole Sonnert, PhD, is an associate on the engineered biology investment team at Playground, where she identifies groundbreaking companies and helps support portfolio companies. Nicole holds undergraduate degrees from Mount Holyoke College and Dartmouth College in Biomathematics and Chemical Engineering, and a PhD from Yale University in Microbiology. Prior to joining Playground, she worked at Adimab, L.L.C. as part of the antibody discovery team. Playground Global led Infinimmune’s seed round in 2022.