The announcement highlights the growing race among scientific institutions and pharmaceutical companies to use advanced AI systems to solve some of biology’s most difficult problems, particularly the challenge of designing proteins that can function safely and effectively inside the human body.
Proteins are often described as the molecular engines of life. They help build tissues, regulate immune responses, transport nutrients, and generate energy throughout the body. Despite decades of scientific progress, creating stable synthetic proteins for therapies has remained a difficult and time-consuming process.
Biohub said its newly released system is powered by the fourth generation of Evolutionary Scale Modeling, commonly known as ESM. The technology studies billions of protein sequences shaped by evolution over millions of years, allowing the AI to learn patterns in how proteins behave, interact, and mutate.
Researchers believe this could dramatically reduce the time needed to identify promising drug candidates for diseases such as cancer and autoimmune disorders.
“We've verified the model's ability and validated many of its predictions in both immune diseases and cancer cases,” Priscilla Chan told Reuters. “It is very promising. We are hopeful that once these models are released, others will quickly adopt them to tackle some of the problems that they see in the lab.”
The launch comes as pharmaceutical companies increasingly invest in artificial intelligence to cut research costs and improve efficiency. AI-driven drug development has become one of the fastest-growing areas in biotechnology, with companies betting that machine learning can shorten years of laboratory work into months.
According to Biohub, the new platform combines several open-source AI models that collectively improve scientists’ understanding of protein behavior and their ability to engineer entirely new proteins.
In early laboratory tests, Biohub researchers said they successfully designed protein binders targeting cancer and immune-related diseases. Those binders reportedly reactivated immune cells, a breakthrough that could contribute to future immunotherapy treatments.
Biohub’s Head of Science, Alex Rives, said accessibility is a key part of the project’s mission.
“We're partnering with a number of different organizations that provide biological analysis platforms, and the models will be available there,” Rives told Reuters. “But we also have a biohub.ai platform, enabling people to use the models on our servers. We will be providing compute credits for that purpose to researchers.”
The AI models are also expected to be distributed through major scientific computing platforms, including Amazon Web Services’ Bio Discovery tools and SandboxAQ, broadening access for researchers around the world.
Biohub has rapidly expanded its role in biomedical research in recent years. The organization, originally established under the Chan Zuckerberg Initiative in 2015, consolidated its scientific operations under the Biohub brand in late 2025, including the acquisition of AI-biology startup EvolutionaryScale.
Since launching their philanthropic efforts, Zuckerberg and Chan have committed more than $7 billion toward charitable causes and previously pledged to donate 99 percent of their Meta shares over their lifetimes, with much of that funding expected to support Biohub’s long-term scientific mission.
Industry analysts say projects like Biohub’s protein world model could mark a turning point in medicine, where AI systems move beyond data analysis and begin actively designing biological solutions for some of humanity’s most complex diseases.