ESM3 open small (esm3-sm-open-v1)
Release in the ESM (EvolutionaryScale and Biohub protein models) family · version esm3-sm-open-v1
Maintained by Biohub (Chan Zuckerberg Biohub, Inc.)14, EvolutionaryScale (original developer; now part of Biohub)57
The openly released small ESM3 model, which EvolutionaryScale describes as a 1.4B-parameter open model. ESM3 is a generative masked language model that takes partial prompts across protein sequence, structure, and function tracks and predicts all three. EvolutionaryScale released it in June 2024; Biohub now distributes it under the MIT license.531
- Model hub: Model card (Hugging Face) (external site: huggingface.co)
- Repository: GitHub repository (Biohub/esm) (external site: github.com)
- Documentation: ESM3 README (external site: github.com)
- Paper: ESM3 paper (Science, doi:10.1126/science.ads0018) (external site: science.org)
Availability and license
Overall availability
Downloadable from Hugging Face without gating. Biohub's README asks users to follow the Biohub Acceptable Use Policy.123
Availability is separate from permission: read the license before using or redistributing.
The model card's text (its metadata has no license field) and the ESM3 README place the model under the MIT license, and the LICENSE file names Chan Zuckerberg Biohub, Inc. as copyright holder (2026). The EvolutionaryScale site still hosts an EvolutionaryScale, PBC Community License Agreement covering the ESM-3 open model code and weights that limits use to non-commercial purposes; the current Biohub materials state MIT, and the sources reviewed do not say when the terms changed. Biohub's README also asks users to follow its Acceptable Use Policy.12346
Model-disclosure tier
The model parameters for this release can be downloaded by the public. License terms may still restrict use, redistribution, or commercial use.
Public materials checklist
| Item | Status | Notes and evidence |
|---|---|---|
| WeightsCan the general public download the model parameters for this release? | Public | Downloadable from Hugging Face without gating.12 |
| Inference codeIs code for running the model published? | Public | The esm package (pip install esm) loads the model locally; the ESM3 README gives an example.31 |
| Training codeIs the code used to train the model published? | Unknown | The public repository covers local inference and API clients; no ESM3 training code was identified in the pages reviewed. |
| Training-data informationPublic = the training data itself can be obtained. Partial = composition or sources are documented without full access. | Partial | The card describes training on 2.78 billion natural proteins, augmented to 3.15 billion sequences, 236 million structures, and 539 million function-annotated proteins (771 billion tokens), with viral proteins and some USDA Select Agent proteins removed. The training set itself was not found as a download.1 |
| Training recipeAre the training configuration and procedure documented in enough detail to follow? | Unknown | The ESM3 paper in Science describes the model; it was not reviewed for this record. |
| Evaluation materialsPublic = evaluation code or prompts that let others re-run the evaluations are published. Partial = results only. | Unknown | Not assessed. |
What it is useful for
Run and use notes
- The ESM3 README loads the model with ESM3.from_pretrained("esm3-sm-open-v1") and runs it on "cuda" or "cpu".3
Organization context
Provenance and derivatives
Other releases in the ESM (EvolutionaryScale and Biohub protein models) family
- ESMC 6B (ESM Cambrian)Model-disclosure tier (USASI rubric v0.1): Open-weight
ESM (EvolutionaryScale and Biohub protein models) family overview
U.S. eligibility
Eligible · basis: U.S.-governed project
Developed by EvolutionaryScale, PBC (New York address in its former terms of use, effective January 15, 2025) and now published by Biohub, whose repository license names Chan Zuckerberg Biohub, Inc. The IRS lists Chan Zuckerberg Biohub Inc in Redwood City, California as a 501(c)(3) organization.8549
Sources
This listing is not an endorsement, a safety assessment, or a federal approval.