Evo 2 by Arc Institute: status and timeline

Evo 2

Evo 2 is a large AI foundation model trained on the genetic sequences of over 128,000 species. It can identify patterns in genomes, predict mutations and design entire genomes at bacterial scale.

DeployStage 4 of 5

Released Feb 2025 via NVIDIA BioNeMo; trained on 9.3T nucleotides from 128K genomes.

Updated 19 Feb 2025Checked 10 Oct0 updates this week

Milestones

Next · Validation studies published

Evo 2 released via NVIDIA BioNeMo19 Feb 2025Complete.
Evo Designer interface launched19 Feb 2025Complete.
Validation studies publishedTarget 2026Current milestone.

Most important updates

Arc Institute and NVIDIA release Evo 2, the largest AI model for biology, trained on 9.3 trillion nucleotides19 Feb 2025

Upcoming

2026Validation studies published (next)

Current obstacles

Validating genome designsEvo 2 can design new sequences, but testing them in living organisms takes months or years; computational validation is incomplete.
Rare organism dataMany organisms lack sequenced genomes, so Evo 2's predictions for them are weaker than for humans or common microbes.

Physics limits

Sequence length vs computeProcessing a million-nucleotide sequence requires significant compute; longer sequences (whole genomes) are harder and slower to analyze.
Prediction accuracy on novel designsEvo 2 is trained on natural sequences; its predictions are more reliable for small mutations in known organisms than for entirely new designs.

How it works

3 parts

Foundation model

Pattern matching at scale

Evo 2 learns patterns from vast genomic data, then applies them to predict how mutations affect genes or design new ones.

StripedHyena 2

Custom architecture

A specialized AI architecture co-developed with OpenAI co-founder Greg Brockman, designed specifically for long genetic sequences.

Public availability

Via NVIDIA BioNeMo

Released for noncommercial use through NVIDIA's cloud platform, with a web interface (Evo Designer) for researchers without ML expertise.

Spec sheet

Spec	Evo 2
Training data	9.3 trillion nucleotides from 128,000+ genomes^R (reported)
Max sequence length	1 million nucleotides^R (reported)
Training compute	2,000+ NVIDIA H100 GPUs via DGX Cloud^R (reported)
Organisms covered	Bacteria, archaea, phage, human, plants and eukaryotes^R (reported)

^R reported by the company

Update log

1 update

Wed 19 Feb 2025

Major: Arc Institute and NVIDIA release Evo 2, the largest AI model for biology, trained on 9.3 trillion nucleotidesBlogResearch

About Arc Institute

The team behind Evo 2

Arc Institute

Evo DNA AI, virtual cell, bridge editing

Nonprofit research institute in Palo Alto founded in 2021 by Patrick Hsu, Silvana Konermann and Patrick Collison, with Stanford, Berkeley and UCSF. It funds scientists without grant applications and builds shared tools like the Evo DNA models.

Evo 2 (Feb 2025, with NVIDIA) learned from DNA of all known life and can predict effects of mutations and write new sequences.
In 2024 Arc researchers showed 'bridge RNAs' that guide enzymes to insert, delete or flip DNA.
The second Virtual Cell Challenge launched on 20 Aug 2026.

arcinstitute.org (opens arcinstitute.org)Company profile →

Founded: 20215 yrs
Headquarters: United States
Status: Private
Valuation: private
Works in: BiotechSynthetic biologyAI
Coverage: 4 programs · 6 updateslatest 20 Aug 2026checked 25 Sep
Partners: NVIDIA (opens nvidia.com)Stanford University (opens stanford.edu)
People: Patrick Hsu (opens arcinstitute.org)Co-founder and core investigatorSilvana Konermann (opens arcinstitute.org)Co-founder and executive directorPatrick Collison (opens en.wikipedia.org)Co-founder

Spec

Evo 2

Training data

9.3 trillion nucleotides from 128,000+ genomes^R (reported)

Max sequence length

1 million nucleotides^R (reported)

Training compute

2,000+ NVIDIA H100 GPUs via DGX Cloud^R (reported)

Organisms covered

Bacteria, archaea, phage, human, plants and eukaryotes^R (reported)