MIT and Recursion Unveil Boltz-2: A Breakthrough Open-Source AI Model for Drug Discovery

In a major leap for computational drug discovery, researchers from the Massachusetts Institute of Technology (MIT) and biotech innovator Recursion (NASDAQ: RXRX) have announced the release of Boltz-2, a cutting-edge AI model designed to predict biomolecular structures and binding affinities with unprecedented speed and precision. This next-generation model, now open-sourced for both academic and commercial use, dramatically advances virtual screening capabilities and sets a new standard in AI-driven drug discovery.

Trained using Recursion's NVIDIA-powered supercomputer, BioHive-2, Boltz-2 is the first open-source biomolecular co-folding model that can jointly predict 3D complex structures and molecular binding affinity. In standard benchmarks, Boltz-2 approaches the accuracy of free energy perturbation (FEP), a gold-standard physics-based method, but performs up to 1,000 times faster, significantly lowering the cost and time required for large-scale molecular screening.

"Accurately predicting how strongly molecules bind has been a long-standing challenge in drug discovery--one that required novel machine learning and computer science techniques to address," said Regina Barzilay, MIT School of Engineering Distinguished Professor for AI and Health, AI faculty lead at Jameel Clinic and CSAIL principal investigator. "Boltz-2 not only addresses this crucial problem but also helps scientists uncover new biological insights and ask questions they couldnt before with standard approaches that are more computationally intensive. Because Boltz-2 is open-source, including its training code, scientists can easily adapt it for specific types of molecules, making it even more powerful as a tool to accelerate discovery.

Built on the foundation of its predecessor, Boltz-1, and inspired by models like AlphaFold3, Boltz-2 distinguishes itself by combining advanced machine learning techniques with expanded datasets, including molecular dynamics simulations and roughly 5 million binding affinity measurements. The model also incorporates Boltz-steering to enhance physical realism, offering greater control and customization options for users.

Among its key technical differentiators:

Near-FEP Accuracy: Delivers FEP+ benchmark-level performance with far less compute time.

Best-in-Class Benchmarks: Outperforms all CASP16 affinity challenge participants.

Joint Prediction Capabilities: Models 3D protein-ligand complexes while simultaneously estimating dynamics like B-factors.

Customizable Outputs: Supports template, contact, and method conditioning for targeted predictions.

Boltz-2 is released under the permissive MIT license, and includes the full model weights and training pipeline, underscoring a joint commitment by MIT and Recursion to foster innovation through transparency and accessibility.

The post MIT and Recursion Unveil Boltz-2: A Breakthrough Open-Source AI Model for Drug Discovery appeared first on PRISM MarketView.

COMTEX_466143566/2927/2025-06-06T11:31:23