NetGoodIndexSubmit a correction

Benefit Ledger · verified · Biology

Boltz-1 releases an MIT-licensed AlphaFold 3–level complex predictor

MIT Jameel Clinic researchers released Boltz-1 in November 2024 as a fully open biomolecular complex model with training code, claiming AF3-level accuracy and an inference-time steering method.

19 Nov 2024Tier 2 NotableMethodology 0.1

Current score

+0.52

3 base · Notable (tier 2 of 5, 3 pts)
× 0.9000 attribution · Primary causal contribution
× 0.5000 evidence · External expert evaluation
× 0.7000 realization · Independently validated or deployed
× 0.5500 durability
Event-level product before credit split: 0.52

Notable democratization of AF3-class modeling (tier 2). High attribution. Preprint plus rapid community use (0.50). Realization is widely installed software. Medium durability as versions increment.

What happened

The bioRxiv paper described architecture and data-processing choices aimed at matching commercial AF3-class accuracy while publishing weights, training code, and datasets under MIT. Boltz-steering was proposed to reduce implausible poses. Subsequent Boltz versions followed. Independent benchmarks generally treat Boltz as a leading open alternative rather than a proven universal winner.

Model attribution

+0.52

Boltz

Open-source biomolecular complex predictor.

The released model is the contribution.

Attribution 0.9000 · Credit share 100% · Massachusetts Institute of Technology

Claims

  • Boltz-1 was released with open training and inference code and reported AF3-comparable complex accuracy.

    outcome · supported

Sources

primary sources

Secondary domains: Computer Science

Revision history

  • 13 Sep 2026 · 0.00 → 0.52

    Initial adjudicated seed score under methodology 0.1.