The Model that Taught Itself Antibodies
ESMfold acts as a protein world model, generating antibodies as an emergent capability without explicit antibody training.
“We didn't design a model for antibodies. We just designed a model that could understand proteins and you kind of get protein design as an emergent property.”
Meta's ESMfold folded over 1.1 billion proteins and emerged as a general-purpose protein world model capable of designing novel antibodies without being explicitly trained to do so. The key insight is that emergent protein design capabilities arise from learning protein biology at scale, collapsing laboratory screening of millions of antibodies into a compute problem. This is a meaningful signal for AI-driven drug discovery and the broader 'world model' framing increasingly applied to scientific domains.