evo-design/evo

Health v2 · maintenance only

Adoption not part of the score

1576 stars · 189 forks observed · 2026-10-13

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

Evo is a 7-billion-parameter DNA foundation model built on the StripedHyena architecture, trained on ~300 billion tokens of prokaryotic whole-genome data for single-nucleotide resolution sequence modeling and generation. This repository provides the Python code and pretrained checkpoints (8k and 131k context, plus CRISPR and transposon finetunes) for inference and genome-scale sequence design.

Use cases

• generate synthetic dna sequences with a genomic language model
• model long dna contexts at single-nucleotide resolution
• design crispr-cas systems with a pretrained model
• generate functional de novo genes
• finetune a genome-scale model on molecular biology tasks
• explore the opengenome dataset for dna pretraining

When to choose

• you need long-context dna sequence modeling or generation in python
• you want pretrained genomic foundation model checkpoints including crispr or transposon variants
• you are doing research in generative genomics or synthetic dna design

When to avoid

• you need a general-purpose protein or rna model rather than dna
• you lack gpu resources for 7b-parameter inference
• you want a production-ready bioinformatics pipeline rather than a research codebase

Facets

library · maturity active

machine-learning deep-learning llm-inference data-generation artificial-intelligence machine-learning bioinformatics large-language-models python dna-foundation-model genomics stripedhyena long-context sequence-modeling biological-design huggingface gpu linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="evo-design/evo")