Science & Data

Boltz

AlphaFold3-class prediction, MIT-licensed

Open ✓Model

What it is

A family of open biomolecular interaction models. Boltz-1 was the first fully open-source structure predictor to approach AlphaFold3 accuracy; Boltz-2 jointly predicts complex structure and binding affinity, claimed to be the first deep-learning model approaching physics-based FEP accuracy while running roughly 1000x faster.

Why it's interesting

A fully MIT-licensed alternative — code and weights — to the restrictively licensed AlphaFold3, with commercial use explicitly permitted. Boltz-2's affinity prediction targets the expensive front end of drug discovery, making it a landmark open release for biology and pharma.

Use cases

  • Protein and complex structure prediction
  • Binding-affinity screening
  • Drug-discovery research

Who it's for

Computational biologists, drug-discovery researchers, bioinformaticians

Setup

Moderate. Python, pip install boltz[cuda]; an NVIDIA GPU is recommended, CPU works but is much slower

Limitations & cautions

Some Boltz-2 training and eval code was still landing at last release, affinity numbers are estimates rather than lab measurements, and large complexes are memory-hungry.

Editorial takeaway

The most consequential open release on this list for people who don't care about AI at all — they care about drugs, and this changes their cost curve.

Related & alternatives