Long-range genomic foundation model handling context lengths up to 1 million base pairs
HyenaDNA uses sub-quadratic Hyena operators to process up to 1 million base pairs at single-nucleotide resolution, unlocking long-range regulatory element and chromatin interaction modeling.
Sub-quadratic scaling allowing context lengths up to 1,000,000 base pairsSingle-nucleotide resolution without k-mer tokenization artifactsTrained on the human reference genome (T2T-CHM13)State-of-the-art benchmarks across the Genomic Benchmarks suite
HyenaDNA uses sub-quadratic Hyena operators to process up to 1 million base pairs at single-nucleotide resolution, unlocking long-range regulatory element and chromatin interaction modeling.
Information checked against an official source; not a hands-on test. Source · Last reviewed: 20/09/2026, 11:09:40
Key Features
Sub-quadratic scaling allowing context lengths up to 1,000,000 base pairs
Single-nucleotide resolution without k-mer tokenization artifacts
Trained on the human reference genome (T2T-CHM13)
State-of-the-art benchmarks across the Genomic Benchmarks suite
Academic Context & Research Evidence
Biological & Workflow Fit
Biological Application
Enhancer-promoter interaction prediction, chromatin profile modeling, and splicing regulation
Research Workflow
Input long DNA sequence -> predict regulatory activity, chromatin marks, and variant effects
Use this citation format when referencing HyenaDNA in scientific publications and benchmark papers.
@software{hyenadna_2026,
title = {{HyenaDNA}},
author = {{Stanford University (Hazy Research)}},
year = {2026},
url = {https://github.com/HazyResearch/hyena-dna},
note = {Indexed on aibioatlas - AI for Biology and Drug Discovery}
}
Peer-Reviewed Literature & Preprints
Live scientific citations streamed from Europe PMC and PubMed for HyenaDNA.
⏳ Fetching real-time literature from Europe PMC & PubMed...
Technical / Product Information
Missing values mean the catalog has no recorded information. They do not mean a feature is absent.
Entry typeAI Model
Access modeOpen Source
AI roleLong-Range Sequence Modeling
Input dataNot recorded
Output dataNot recorded
Licence conditionsApache 2.0
Commercial eligibilityOpen source release
Compute requirements1x GPU (A100 or H100 recommended for 1M context)
ValidationNot recorded
TypeLong-context genomic foundation model
Intended useNot recorded
CompatibilityNot recorded
ManufacturerStanford University
Biological applicationEnhancer-promoter interaction prediction, chromatin profile modeling, and splicing regulation
Research workflowInput long DNA sequence -> predict regulatory activity, chromatin marks, and variant effects
Evidence levelPeer-reviewed research (NeurIPS 2023)