# Evo 2 — weights, license and what is actually open

A biological foundation model trained across genomes for sequence understanding and design, released with code, weights, and OpenGenome2 data.

Canonical: https://brightaifuture.com/open-models/evo
Format: model-family
Source publication: Not established
Bright publication: 2026-09-19
Substantive update: None recorded
Evidence and review: AI-assisted primary-source check on 2026-09-19; no independent model reproduction or license certification. The peer-reviewed manuscript includes collaborator evaluations; no claim of clinical validation is made.

## Documented release

Evo 2 7B and 40B

Organization: Arc Institute and collaborators

Evo 2 release: 2025-02-19 (day precision). Source: https://arcinstitute.org/news/evo2

Parameters: 7B and 40B

Architecture: StripedHyena 2 sequence model for million-token genomic contexts.

Modalities: DNA sequence

Context: Up to 1M nucleotides

## License and commercial use

Apache-2.0 for code and model artifacts; dataset terms must be checked separately

Model code and weights are permissively licensed; biological datasets and downstream applications require separate diligence.

## What is open

Weights: available. 7B and 40B model parameters are released. Source: https://arcinstitute.org/manuscripts/Evo2.pdf

Architecture: available. The paper and code describe StripedHyena 2. Source: https://arcinstitute.org/manuscripts/Evo2.pdf

Inference code: available. Official inference and scoring code is public. Source: https://github.com/ArcInstitute/evo2

Training code: available. Pretraining, midtraining and fine-tuning code is released separately in Savanna, as linked by the Evo 2 paper. Source: https://github.com/zymrael/savanna

Training recipe: available. The paper documents model and training methodology. Source: https://arcinstitute.org/manuscripts/Evo2.pdf

Data information: available. OpenGenome2 construction is documented. Source: https://arcinstitute.org/manuscripts/Evo2.pdf

Training data: available. OpenGenome2 is released for research use under its stated terms. Source: https://arcinstitute.org/manuscripts/Evo2.pdf

Evaluation: available. The paper publishes biological evaluations and methods. Source: https://arcinstitute.org/manuscripts/Evo2.pdf

Commercial use: partial. Code and weights are permissive, while dataset and application constraints are not reducible to one permission. Source: https://arcinstitute.org/manuscripts/Evo2.pdf

## Uses and strengths described in sources

Genomic sequence modeling

Variant scoring

Long biological context

Open data and code

## Hardware and quantization

The 40B model is server-class; the 7B model broadens access but still depends on context length and precision.

No canonical quantized release is asserted here

## Independent evidence

The peer-reviewed manuscript includes collaborator evaluations; no claim of clinical validation is made.

## Limitations

Research model, not a clinical diagnostic

Generated biological sequences require domain safety review

## Provenance and history

{}

## Original sources

- [Arc Institute Evo 2 announcement](https://arcinstitute.org/news/evo2)
- [Genome modeling and design across all domains of life with Evo 2](https://arcinstitute.org/manuscripts/Evo2.pdf)
- [Savanna training code linked by Evo 2](https://github.com/zymrael/savanna)
- [Evo 2 repository](https://github.com/ArcInstitute/evo2)

## Continue exploring

- [Open Models](https://brightaifuture.com/open-models)
- [Open intelligence thread](https://brightaifuture.com/threads/open)