Gabriele Corso is a co-founder of Boltz, a public benefit corporation that develops open models and agents for AI-driven drug discovery. Corso co-authored the Boltz-1 and Boltz-2 biomolecular models, which began as research at MIT, and is first author of DiffDock, a diffusion model for molecular docking.123
At a glance
- Current role
- Co-founder
- Company
- Boltz
- Sector
- Company
- Fields
- AI for science
- Work listed
- 3
- Sources cited
- 7
- Profile reviewed
Research
Corso's research applies generative models to molecular structure. DiffDock, posted in October 2022 with Hannes Stärk, Bowen Jing, Regina Barzilay and Tommi Jaakkola, treats docking a small molecule to a protein as a diffusion process over the molecule's position, orientation and torsion angles.3
Boltz models
Boltz is a family of open models for predicting biomolecular interactions, released with code and weights under the MIT license. The Boltz-1 technical report appeared in November 2024, and Boltz-2, which jointly models complex structures and binding affinities, followed in June 2025.256
Boltz PBC
The Boltz models began as research at MIT. In January 2026, Gabriele Corso, Jeremy Wohlwend and Saro Passaro announced that they had founded Boltz as a public benefit corporation to keep releasing open models while building products for drug discovery.1
The launch introduced a preview of Boltz Lab, a platform for running the models and new small-molecule and protein design agents, along with a partnership with Pfizer.4
Work and contributions
Boltz-2 Open weights
Co-author of Boltz-2 and Boltz-1, open biomolecular interaction models whose code and weights are released under the MIT license.256
DiffDock
First author of DiffDock (2022), which frames molecular docking as a generative modeling problem solved with a diffusion process over ligand poses.3
Boltz-2 (boltz2_conf and boltz2_aff checkpoints) Open weights
Second author of the Boltz-2 paper, which the Boltz repository README asks users of the code and models to cite.76
Timeline
- First-authored the DiffDock paper on diffusion models for molecular docking.3
- Posted the Boltz-1 technical report on democratizing biomolecular interaction modeling.5
- Posted the Boltz-2 technical report on structure and binding affinity prediction.6
- Announced the founding of Boltz as a public benefit corporation with co-founders.14
- Boltz launched a preview of the Boltz Lab platform and its first design agents.4
Where they work
- Boltz 3 people in the wiki
Related people
- Jeremy WohlwendAlso at Boltz
- Saro PassaroAlso at Boltz
- Jong Youl ChoiAI for science
- Alexandre GramfortAI for science
- Josef SivicAI for science
- Priscilla ChanAI for science
- Stephen K. StreifferAI for science
- Massimiliano Lupo PasiniAI for science
- Aditya GroverAI for science
- Ali MadaniAI for science
Suggest a correction or removal
Profiles cover public professional work only, and each statement cites a source. If something is out of date, missing or wrong, tell us and link a source. People can also ask for their profile to be corrected or removed.
References
7 sources · profile reviewed Oct 21, 2026 · data updated Oct 22, 2026
- Our Vision: The Future We Are Building at Boltz
- jwohlwend/boltz README (models, MIT license, citations)
- DiffDock: Diffusion Steps, Twists, and Turns for Molecular Docking (arXiv 2210.01776)
- Announcing Boltz: A Public Benefit Corporation for AI-Driven Drug Discovery
- Boltz-1: Democratizing Biomolecular Interaction Modeling (bioRxiv)
- Boltz-2: Towards Accurate and Efficient Binding Affinity Prediction (bioRxiv)
- jwohlwend/boltz README (raw)