USASI
Model family

ESM (EvolutionaryScale and Biohub protein models)

Family overview — summarizes releases; licenses and availability belong to each release.

Maintained by Biohub (Chan Zuckerberg Biohub, Inc.)12, EvolutionaryScale (original developer; now part of Biohub)567

Protein language models developed by EvolutionaryScale and, since its team joined Biohub in November 2025, released by Biohub. The line includes ESM3 (June 2024), a generative model that reasons over protein sequence, structure, and function, with a largest version of 98B parameters and a small open model; ESM C (ESM Cambrian, December 2024), representation models at 300M, 600M, and 6B parameters; and, from May 2026, ESMFold2 structure prediction models and sparse autoencoders built on ESM C.5631109

Last reviewedEntry updated Documented release Jun 25, 2024

Releases assessed

Each release is assessed on its own. This family has no family-wide openness label.
Releases in the ESM (EvolutionaryScale and Biohub protein models) family
ReleaseWeightsTier (USASI rubric v0.1)LicenseReleased
ESM3 open small (esm3-sm-open-v1)Weights: PublicModel-disclosure tier (USASI rubric v0.1): Open-weightMITJun 25, 2024
ESMC 6B (ESM Cambrian)Weights: PublicModel-disclosure tier (USASI rubric v0.1): Open-weightMITMay 2026

What it is useful for

Protein representation learning and fine-tuning (ESM C), protein generation from partial sequence, structure, and function prompts (ESM3), and structure prediction for proteins and other biomolecules (ESMFold2).314

Run and use notes

Documented facts only. No hardware or performance claims are made without a cited source and stated assumptions.
  • The ESM3 README lists medium (7B) and large (98B) ESM3 models that are accessed through Biohub, with only the esm3-sm-open-v1 weights on Hugging Face; the hosted API moved from forge.evolutionaryscale.ai to biohub.ai, and the esm SDK still uses Forge class names.31

Organization context

Provenance and derivatives

ESM3 was trained on 2.78 billion natural proteins (771 billion unique tokens at its largest scale); ESM C was trained on protein sequences from UniRef, MGnify, and the Joint Genome Institute. ESMFold2 is built on ESMC 6B language model embeddings.361

U.S. eligibility

Project eligibility rests on documented governing or maintaining entities, not on contributors.

Eligible · basis: U.S.-governed project

The models were developed by EvolutionaryScale, PBC, whose former terms of use (effective January 15, 2025) gave a New York address, and are now published by Biohub; the repository license names Chan Zuckerberg Biohub, Inc. as copyright holder. The IRS lists Chan Zuckerberg Biohub Inc in Redwood City, California as a 501(c)(3) organization. Eligibility of each release is assessed on the release record.82111

Assessed Sep 29, 2026

Sources

  1. 1.
  2. 2.
    Biohub/esm LICENSE.md (external site: raw.githubusercontent.com)

    Chan Zuckerberg Biohub, Inc. · License · accessed Sep 29, 2026

  3. 3.
  4. 4.
    biohub/ESMFold2 model card (external site: huggingface.co)

    Biohub · Model card · accessed Sep 29, 2026

  5. 5.
    ESM3: Simulating 500 million years of evolution with a language model (external site: evolutionaryscale.ai)

    EvolutionaryScale · Announcement · published Jun 25, 2024 · accessed Sep 29, 2026

  6. 6.
    ESM Cambrian: Revealing the mysteries of proteins with unsupervised learning (external site: evolutionaryscale.ai)

    EvolutionaryScale · Announcement · published Dec 4, 2024 · accessed Sep 29, 2026

  7. 7.
    EvolutionaryScale (home page) (external site: evolutionaryscale.ai)

    EvolutionaryScale · Official page · accessed Sep 29, 2026

  8. 8.
    Terms of Use (historical) | EvolutionaryScale (external site: evolutionaryscale.ai)

    EvolutionaryScale · Official page · accessed Sep 29, 2026

  9. 9.
  10. 10.
    Biohub releases a world model of protein biology (external site: biohub.org)

    Biohub · Announcement · published May 27, 2026 · accessed Sep 29, 2026

  11. 11.

This listing is not an endorsement, a safety assessment, or a federal approval.

Support Us

Help keep USASI useful.

Optional. No USASI account required. Payment takes place on the linked provider’s website (Buy Me a Coffee).

About supporting this project