Biohub Launches 6.8 Billion-Protein AI Model to Speed Drug Discovery
Updated
Updated · pharmaphorum · May 27
Biohub Launches 6.8 Billion-Protein AI Model to Speed Drug Discovery
12 articles · Updated · pharmaphorum · May 27
Biohub released a free AI "world model" of protein biology that it says can predict structures, map protein patterns and design therapeutic molecules far faster than conventional methods.
Built on an atlas of 6.8 billion proteins and 1.1 billion structures, the system combines Biohub's ESMC language model with ESMFold2 to model folding, binding and protein interactions.
Biohub said the platform can cut discovery of new protein binders from months or years to days or even hours, positioning it as an open alternative to systems such as DeepMind's AlphaFold.
Tests on five cancer and immunology targets — including EGFR, PD-L1 and CTLA-4 — produced hit rates of 36%-88% for minibinders and 15%-29% for antibody modalities.
Backed by Priscilla Chan and Mark Zuckerberg, the nonprofit said making the model openly available is meant to broaden access to protein design tools and speed personalized therapy research.
Is the new AI drug discovery tool building on a foundation of flawed biological data?
As open-source AI rivals private tech, what is the new business model for drug discovery?
With AI automating protein design, what is the new role for human scientists in drug discovery?
Biohub’s AI Protein World Model: Democratizing Protein Science and Advancing Toward Virtual Cell Simulations
Overview
In late 2025 or early 2026, Biohub unveiled its AI-powered world model for protein biology, marking a pivotal moment in digital biology. This launch represents a major leap forward in understanding the fundamental building blocks of life and accelerates the quest for new therapies. Driven by Biohub's belief that open science accelerates discovery, the model's components are freely available, empowering researchers globally. This accessibility fosters rapid progress toward personalized cures that effectively target the specific biological mechanisms driving individual patients' diseases, making advanced protein science more accessible and impactful for the scientific community.