Skip to content
← Models
Open weightsSalesforce Research/10 trustMetrics-checked

ProGen2

Autoregressive protein language models for generating functional protein sequences.

Visit official page →

About

ProGen2 is Salesforce Research's family of autoregressive protein language models, first released in 2022 and published in Cell Systems in 2023. The family spans 151M to 6.4B parameters, trained on natural proteins from UniRef90, the BFD metagenomic set, and antibody sequences.

Trained with next-residue prediction, the models can sample novel protein sequences from a prompt and score existing sequences as a zero-shot proxy for fitness. Checkpoints are BSD-3-Clause, free for academic and commercial use. Release timing below is era-level; the paper is the canonical reference.

Specs

Context window

Pricing

Free quota

API access

Weights

Open weights

Released

January 1, 2022

Benchmarks

Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.

Use cases

protein-designdrug-discoveryresearch

Trust score — 7/10

/10 trustMetrics-checked

Scored 7/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 110; Trending score 74/100 from researcher assessment.

Sources