Evo2

License: Evo2 is open source and free for academic and commercial use under an Apache-2.0 license. Please refer to the license for full terms.

Proto is not affiliated with Arc Institute. This toolkit is open source and builds on the implementation produced by this organization. Product names, logos, and trademarks are the property of their respective owners.

GitHub GitHub Publication Publication Preprint Preprint Cite Cite Tool Source Tool Source Open as Notebook Open as Notebook Run on Modal Run on Modal

arcinstitute/evo2

View repo

Genome modelling and design across all domains of life with Evo 2

Garyk Brixi, Matthew G Durrant, … Brian L Hie

Nature (2026)

Read paper

Genome modeling and design across all domains of life with Evo 2

G. Brixi, Matthew G. Durrant, … Brian L. Hie

bioRxiv (2025)

Read preprint

@articleBrixi2026-jn

@ARTICLE{Brixi2026-jn,
  title     = "Genome modelling and design across all domains of life with Evo 2",
  author    = "Brixi, Garyk and Durrant, Matthew G and Ku, Jerome and
               Naghipourfar, Mohsen and Poli, Michael and Sun, Gwanggyu and
               Brockman, Greg and Chang, Daniel and Fanton, Alison and Gonzalez,
               Gabriel A and King, Samuel H and Li, David B and Merchant, Aditi
               T and Nguyen, Eric and Ricci-Tam, Chiara and Romero, David W and
               Schmok, Jonathan C and Taghibakhshi, Ali and Vorontsov, Anton and
               Yang, Brandon and Deng, Myra and Gorton, Liv and Nguyen, Nam and
               Wang, Nicholas K and Pearce, Michael T and Simon, Elana and
               Adams, Etowah and Amador, Zachary J and Ashley, Euan A and
               Baccus, Stephen A and Dai, Haoyu and Dillmann, Steven and Ermon,
               Stefano and Guo, Daniel and Herschl, Michael H and Ilango, Rajesh
               and Janik, Ken and Lu, Amy X and Mehta, Reshma and Mofrad,
               Mohammad R K and Ng, Madelena Y and Pannu, Jaspreet and R{\'e},
               Christopher and St John, John and Sullivan, Jeremy and Tey,
               Joseph and Viggiano, Ben and Zhu, Kevin and Zynda, Greg and
               Balsam, Daniel and Collison, Patrick and Costa, Anthony B and
               Hernandez-Boussard, Tina and Ho, Eric and Liu, Ming-Yu and
               McGrath, Thomas and Powell, Kimberly and Pinglay, Sudarshan and
               Burke, Dave P and Goodarzi, Hani and Hsu, Patrick D and Hie,
               Brian L",
  journal   = "Nature",
  publisher = "Springer Science and Business Media LLC",
  pages     = "1--13",
  doi       = "10.1038/s41586-026-10176-5",
  month     =  mar,
  year      =  2026,
  language  = "en"
}

Copy citation

evo-design/proto-tools

View source

Open Notebook

Open notebook

Install MCP Evo2 Sampling Evo2 Scoring

Toolkit contributors

Function	Description
`run_evo2_icl()`	Generate DNA or protein designs from FASTA examples with pretrained Evo 2 (GPU)	Docs Source
`run_evo2_sample()`	Sample DNA sequences using Evo2 language model (GPU)	Docs Source
`run_evo2_score()`	Score DNA sequences using Evo2 language model (GPU)	Docs Source

Background

Evo2 (Brixi et al., 2026) is a DNA language model trained with an autoregressive objective: during training the model learns to predict the next nucleotide given all preceding nucleotides. Training used the OpenGenome2 dataset, which spans bacterial, archaeal, eukaryotic, and phage genomes across all domains of life, so the model is not restricted to any single clade. It is available at several scales, the largest being 40 billion parameters, and uses the StripedHyena 2 architecture, a sequence model that combines convolutional state-space layers with a smaller number of attention layers. This design lets the model process very long stretches of DNA, up to roughly one million nucleotides for the long-context checkpoints, without the memory cost a pure attention model would incur at that length. Several checkpoints are also offered with shorter context windows for lower memory use, and one variant is trained specifically on Microviridae phage genomes. The autoregressive objective yields two capabilities directly. Sampling from the predicted next-nucleotide distributions produces new candidate sequences, and reading off the probabilities the model assigns to an existing sequence gives a likelihood score that reflects how closely the sequence matches the patterns seen during training. Evo2 is the second model in the Evo family; the earlier Evo1 was trained only on prokaryotic and phage genomes, whereas Evo2 extends to eukaryotic genomes and longer context.

Learning Resources

The Illustrated Evo 2 (NVIDIA Research) - a visual walkthrough of the Evo 2 architecture and how the model processes and generates DNA.
Evo 2 Mechanistic Interpretability (Arc Institute) - an interactive look at the internal features Evo 2 learns, built with sparse autoencoders to surface interpretable genomic patterns.

Tools

Evo 2 In-context Design (`evo2-icl`)

In-context learning (ICL) conditions pretrained Evo 2 7B on example DNA without retraining. Supply a FASTA-derived pool of DNA or protein sequences; the tool selects examples, composes DNA prompts, and returns designs in the input alphabet for FASTA export.

API Reference

Source

Input: Evo2ICLInput

enum

default:"dna"

DNA accepts A/C/G/T; protein accepts the 20 standard amino acids.Available options: dna, protein

List[string]

required

Sequences in FASTA record order, without headers. Whitespace is removed and sequences are uppercased.

Source

Config: Evo2ICLConfig

integer

default:"20"

Designs per composed prompt, without a library-level count cap.

integer

default:"16"

Examples selected without replacement per prompt, capped by pool size.

integer

default:"1"

Number of prompt draws, each producing num_sequences designs.

boolean

default:"False"

Use the first n_shots records in order instead of random selection.

integer

default:"42"

Seed for example selection, independent of generation.

string

default:"N"

DNA delimiter preceding each example and ending each prompt.

boolean

default:"False"

Also append the delimiter after every example.

integer

default:"1"

Minimum length required for every example in the pool.

integer

Parallel prefill length; None selects a batch-aware value.

integer

default:"42"

Seed for uniform reverse translation, independent of generation.

integer

default:"200"

New bases (DNA mode) or amino acids (protein mode) per design; the example sequences and separators are excluded from the output.

number

default:"1.0"

Sampling randomness. Lower values favor likely nucleotides; higher values explore more alternatives.

integer

default:"4"

Number of highest-probability tokens considered at each step.

number

default:"0.9"

Cumulative probability cutoff for nucleus sampling.

integer

default:"1"

Number of designs decoded together on the GPU.

integer

default:"0"

Verbosity level (0=quiet, 1=info, 2=debug, 3=raw subprocess stderr). True is coerced to 1 and False to 0.

string

default:"cuda"

Device to run the model on.

integer

default:"14400"

Maximum execution time in seconds.

integer

Generation seed; None draws fresh designs.

Source

Output: Evo2ICLOutput

enum

Alphabet of the returned designs.

List[string]

Corresponding generated DNA for protein results, in matching order.

List[CausalModelSample]

required

Requested-length DNA or protein designs. Protein translation starts at the first generated base; stops remain * and ambiguous codons become X, with warnings. No early stopping is used.

Show CausalModelSample

string

required

The generated sequence.

Applications

Use examples of a sequence family, such as regulatory DNA with a shared experimental activity, to condition candidate generation. The ICL study is Few-shot regulatory sequence design with a genomic language model (manuscript; paper URL coming soon). The HepG2 example pool contains 463 regulatory sequences of 200 bp each.

Usage Tips

Default run. Draw up to 16 examples, join them with a single N separator, and generate 20 designs of 200 bp. Generation always runs to the requested length. Sampling defaults are temperature 1.0, top K 4, and top P 0.9.
Choose the examples. n_shots sets examples per prompt; n_prompts sets the number of prompts, each producing num_sequences designs. Use fixed_examples=True to take the first records in FASTA order instead of sampling.
Protein input and output. Set sequence_type="protein" to supply the 20 standard amino acids. Examples are reverse-translated by uniformly sampling synonymous codons from the standard genetic code, using codon_seed (default 42). Length is measured in amino acids. Outputs include protein designs and matching dna_sequences; stop codons remain * and ambiguous codons become X, with warnings. No start or stop codon is added.
Reproduce a run. example_seed controls example selection, codon_seed controls reverse translation, and seed controls generation. Keep all three seeds and the batch size fixed to reproduce a run; change only seed to sample new designs from the same prompts.

Evo2 Sampling (`evo2-sample`)

Generates DNA sequences by autoregressive sampling. Given one or more prompt sequences in Evo2’s prompt format, the model extends each prompt nucleotide by nucleotide, drawing each new nucleotide from the model’s predicted distribution under the configured temperature, top_k, and top_p settings, until max_new_tokens new nucleotides have been produced or an end-of-sequence token is sampled. A key-value cache makes long generations efficient and can be carried forward to continue a generation.

API Reference

Source

Input: CausalModelSampleInput

List[string]

required

Prompt sequences to condition generation on. Can be provided as a single string or a list of strings.

Source

Config: Evo2SampleConfig

enum

default:"evo2_7b"

Evo2 weights variant.Available options: evo2_7b, evo2_20b, evo2_40b, evo2_7b_base, evo2_40b_base, evo2_1b_base, evo2_7b_262k, evo2_7b_microviridae

string

Override HuggingFace download with a local weights directory.

integer

default:"4"

Limit sampling to the top-k most probable tokens at each step.

integer

default:"32"

Maximum number of new tokens to generate per prompt (excludes prompt).

boolean

default:"True"

Use the model’s per-call KV cache during generation.

integer

Tokens to prefill in parallel before switching to autoregressive prompt forcing.

integer

Maximum sequence length the KV cache will be sized for.

boolean

default:"False"

Filter EOS/PAD bytes from the detokenized output.

boolean

default:"True"

Stop generation when an EOS (id=0) token is sampled.

Evo2KVCacheRef

Worker-local KV cache handle returned by a previous persistent-worker generation call.

boolean

default:"False"

Return worker-local KV cache handles for continued generation.

boolean

default:"False"

Include per-position logits in the output.

integer

default:"0"

Verbosity level (0=quiet, 1=info, 2=debug, 3=raw subprocess stderr). True is coerced to 1 and False to 0.

string

default:"cuda"

Device to run the model on.

integer

default:"1800"

Maximum execution time in seconds. None waits indefinitely.

integer

Random seed. When set, tools run reproducibly up to small GPU float noise (see BaseToolOutput.approx_equal), and the seed participates in cache keys. When None, cacheable seed-sensitive tools skip cache until seeded.

boolean

default:"True"

Include the input prompt at the start of each generated sequence; when False, only newly generated tokens are returned.

number

default:"1.0"

Sampling temperature controlling randomness.

number

default:"1.0"

Nucleus sampling threshold over per-position token probabilities.

integer

default:"4"

Number of sequences to process simultaneously.

Source

Output: Evo2SampleOutput

List[Evo2Sample]

required

One generated DNA sequence per prompt, with its logits and cache handle.

Show Evo2Sample

any

Per-position logits for this sequence (shape: [num_generated_tokens, vocab_size]).

Evo2KVCacheRef

Opaque worker-local cache handle for continued generation, valid only on the worker that produced it.

string

required

The generated DNA sequence.

Applications

This tool produces candidate DNA sequences for downstream design and screening, including genes, regulatory regions, and longer multi-gene segments. Because Evo2 is trained across all domains of life, it can be prompted with eukaryotic as well as prokaryotic and phage context, unlike the prokaryote-and-phage-only Evo1. The prompt sets the biological context for what follows.

Usage Tips

Match the checkpoint to the task. evo2_7b (the default), evo2_20b, and evo2_40b are the 1M-context models in increasing size and capability. The evo2_7b_base, evo2_40b_base, and evo2_1b_base checkpoints are 8K-context counterparts (evo2_1b_base is the smallest); evo2_7b_262k is a 262K-context variant; evo2_7b_microviridae is a 7B model adapted on Microviridae genomes for generating that bacteriophage family.
Prompts use Evo2’s prompt format. Prompt strings follow Evo2’s special tokenization (for example a leading +~ before DNA); see the upstream Evo2 documentation for the conventions.
top_k defaults to 4, the size of the DNA alphabet. It exists mainly to keep generation on the four bases rather than other byte tokens, so it is not the diversity parameter; control diversity with temperature (lower stays near the training distribution, higher explores it) and leave top_p at its default unless you specifically want nucleus sampling.
Output includes the prompt by default. prepend_prompt=True (the default for this toolkit) returns the prompt joined to its continuation; set it False to receive only the newly generated nucleotides.
Prompt length plus max_new_tokens (default 32) must fit the checkpoint’s context window. The model cannot attend beyond that window, so a long prompt directly reduces how much can be generated; pick a longer-context checkpoint when the combined length is large.
stop_at_eos ends generation early when the model emits an end-of-sequence token; set it to False to always produce the full max_new_tokens.
Generated sequences are candidates. Validate them with downstream tools (for example ORF detection, structure prediction, or homology search) before drawing biological conclusions.

Evo2 Scoring (`evo2-score`)

Scores existing DNA sequences under the Evo2 model. For each sequence, it computes the model’s predicted probability of every nucleotide given the preceding nucleotides and aggregates these into a log-likelihood, an average log-likelihood per nucleotide, and a perplexity. Optionally returns the per-position logits, alongside the token vocabulary giving their column order.

API Reference

Source

Input: CausalModelScoringInput

List[string]

required

Sequences to score. Can be provided as a single string or a list of strings.

Source

Config: Evo2ScoringConfig

enum

default:"evo2_7b"

Evo2 weights variant.Available options: evo2_7b, evo2_20b, evo2_40b, evo2_7b_base, evo2_40b_base, evo2_1b_base, evo2_7b_262k, evo2_7b_microviridae

string

Override HuggingFace download with a local weights directory.

boolean

default:"False"

Prepend a beginning-of-sequence token before scoring.

integer

default:"0"

Verbosity level (0=quiet, 1=info, 2=debug, 3=raw subprocess stderr). True is coerced to 1 and False to 0.

string

default:"cuda"

Device to run the model on.

integer

default:"1800"

Maximum execution time in seconds. None waits indefinitely.

integer

default:"8"

Number of sequences to process simultaneously on GPU. Larger batches improve throughput but use more GPU memory; reduce if encountering out-of-memory errors.

boolean

default:"False"

Include per-position logits in the output.

Source

Output: CausalModelScoringOutput

List[CausalModelScoringMetrics]

required

List of scoring outputs, one per input sequence. Each entry is a Metrics subclass with scalar metrics (log_likelihood, avg_log_likelihood, perplexity) and optional per-position _pp-suffixed list extras; logits and vocab are declared fields for raw model outputs.

Show CausalModelScoringMetrics

array

Per-position logits array (seq_len, vocab_size). None unless return_logits=True.

array

Token ordering for logits.

string

Name of the metric that best summarizes the result overall (e.g. "avg_plddt" for AlphaFold2). Used by downstream UI and reporting to pick a headline value.

string

Concrete Metrics subclass tag; enables typed reconstruction after a serialization round-trip.

Metrics (one set per scores item)

Metric	Type	Range	Availability
`log_likelihood`	float	≤ 0.0	always
`avg_log_likelihood`	float	≤ 0.0	always
`perplexity`	float	≥ 1.0	always

Applications

This tool measures how well a DNA sequence matches the patterns the model learned during training across all domains of life. Lower perplexity means the sequence is more consistent with that distribution. Use it to rank or filter candidate sequences (including the output of evo2-sample), to compare variants of a sequence, or to assess sequences from organisms outside the prokaryotic and phage range that Evo1 covers.

Usage Tips

Compare length-normalized scores within one checkpoint. Total log_likelihood scales with sequence length, so use perplexity or avg_log_likelihood when comparing sequences of different lengths. Different checkpoints learn different distributions that are not calibrated to a common scale, so scores from different model_checkpoint values are hard to compare directly; a lower perplexity means the sequence is more consistent with that checkpoint’s training distribution.
return_logits defaults to False. Leave it off unless you need the per-position distributions, since the logits tensor is large (sequence length by a 512-token vocabulary).
prepend_bos adds a beginning-of-sequence token before scoring; leave it False unless matching a specific upstream convention.

Toolkit Notes

These apply to every Evo2 tool in this toolkit (evo2-icl, evo2-sample, evo2-score).

Requires a high-memory GPU; memory scales with model size and context length. The 7B checkpoint needs a high-memory NVIDIA GPU; the 20B and 40B models and the 1M-context checkpoints need substantially more. CPU execution is not practical.
batch_size trades memory for throughput. It sets how many designs (evo2-icl), prompts (evo2-sample), or sequences (evo2-score) are processed per GPU forward pass. Raise it for higher throughput on many short sequences; lower it if generation or scoring runs out of GPU memory.
Trained across all domains of life. Evo2 covers prokaryotic, eukaryotic, archaeal, and phage genomes. For prokaryote-and-phage-only generation with a smaller model, Evo1 is an alternative.

Example notebook: See the full working example for a copy-paste-ready walkthrough.

Infrastructure Guides

The following guides cover how to run tools efficiently and at scale.

Tool Persistence

Keep a tool’s model warm across calls instead of reloading it every invocation.

Device Management

How GPUs are allocated to tools and how to target specific devices.

Parallel Execution

Fan a batch of inputs out across multiple GPUs.

Tools

Guides

Tool Catalog

Concepts

Implementing a Tool

Background

Learning Resources

Tools

Evo 2 In-context Design (`evo2-icl`)

API Reference

Applications

Usage Tips

Evo2 Sampling (`evo2-sample`)

API Reference

Applications

Usage Tips

Evo2 Scoring (`evo2-score`)

API Reference

Applications

Usage Tips

Toolkit Notes

Infrastructure Guides

Tool Persistence

Device Management

Parallel Execution

​Background

​Learning Resources

​Tools

​Evo 2 In-context Design (evo2-icl)

​API Reference

​Applications

​Usage Tips

​Evo2 Sampling (evo2-sample)

​API Reference

​Applications

​Usage Tips

​Evo2 Scoring (evo2-score)

​API Reference

​Applications

​Usage Tips

​Toolkit Notes

​Infrastructure Guides

Tool Persistence

Device Management

Parallel Execution

Background

Learning Resources

Tools

Evo 2 In-context Design (`evo2-icl`)

API Reference

Applications

Usage Tips

Evo2 Sampling (`evo2-sample`)

API Reference

Applications

Usage Tips

Evo2 Scoring (`evo2-score`)

API Reference

Applications

Usage Tips

Toolkit Notes

Infrastructure Guides