Boltz-2
Biomolecular co-folding model predicting structure and binding affinity
About
Boltz-2 is a biomolecular foundation model from MIT CSAIL and Recursion, released on June 6, 2025 under the MIT license. It jointly predicts 3D complex structures and binding affinities in a single pass, approaching the accuracy of physics-based free-energy methods at a fraction of the compute. The team reports it beat every submitted method in the CASP16 binding-affinity sub-track. Unlike AlphaFold 3, the weights and training code are fully open for commercial use.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Team-reported: Pearson correlation 0.62 with true binding affinities on the FEP+ benchmark at 1,000x the speed of physics-based methods; outperformed all submitted methods in the CASP16 affinity sub-track.
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 9/10
Scored 9/10 (verified tier, metrics-v1). Evidence: MIT CSAIL + Recursion release; MIT license confirmed; CASP16 affinity sub-track win is team-reported via MIT release; Fully open weights, code and training details — unusual for this tier. License: MIT (per release announcement — commercial use allowed)..
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 8.0/10
- Maintenance · recency, releases, issue load
- 9.0/10
- Openness · license + docs
- 9.0/10
- Manual review · sanity read + red-flag check
- 9.0/10
- · MIT CSAIL + Recursion release; MIT license confirmed
- · CASP16 affinity sub-track win is team-reported via MIT release
- · Fully open weights, code and training details — unusual for this tier