BOTANIC?
Living Models

Living Models

BOTANIC?

Genomic language models
for land plants.

Model cards Technical report bioRxiv

Model size

318M to 3.2B parameters

Botanic1-S scores 0.758 on the aggregate benchmark across nine task families. PlantCAD2-L scores 0.756, Carbon-8B 0.727, and Evo 2-7B 0.695.

Frozen models · aggregate test score (Sbaltest) · whiskers are 95% CIs

Scores by task and species

Species coverage

320 land plant species

Pre-training spans bryophytes, vascular spore plants, gymnosperms, and flowering plants.

Explore the training species

Zero-shot scoring

Variant prioritisation

Botanic1-XL recovers 49.9% of validated causal variants within the top 1% of candidates (~60 variants out of a median of 5,904 per locus), across 545 studies in 14 species, from the sequence and mutation allele alone.

Explore the variant benchmark

Fine-tuning

Molecular phenotype predictions

Easier and better fine-tuning for promoter and terminator strength, poly(A) sites, lncRNA classification, chromatin accessibility, and TF-family binding.

Full fine-tuning · LoRA · IA³

Methods and results in the report

Evaluations, uncertainty estimates, and species-level results are detailed in the technical report.

Cite

Barozet A., Cabeli V., Ogier du Terrail J., Rukhovich A., Janssoone T., Klajer G., Sheikhitarghi Z., Andrews G., Veran C., Strouk L. BOTANIC-1: a series of long-context plant genomic foundation models in the agentic era. bioRxiv (2026). doi: 10.64898/2026.09.04.749355

@article{Barozet2026.09.04.749355,
    author = {Barozet, Am{\'e}lie and Cabeli, Vincent and Ogier du Terrail, Jean
              and Rukhovich, Alexey and Janssoone, Thomas and Klajer, Gary
              and Sheikhitarghi, Zeinab and Andrews, Gregory and Veran, Cyril
              and Strouk, L{\'e}onard},
    title = {BOTANIC-1: a series of long-context plant genomic foundation
             models in the agentic era},
    journal = {bioRxiv},
    year = {2026},
    elocation-id = {2026.09.04.749355},
    doi = {10.64898/2026.09.04.749355},
    publisher = {Cold Spring Harbor Laboratory},
    URL = {https://www.biorxiv.org/content/early/2026/09/09/2026.09.04.749355},
    eprint = {https://www.biorxiv.org/content/early/2026/09/09/2026.09.04.749355.full.pdf},
}

Model cards contain instructions and tutorials

Score variants

Provide a reference sequence, a variant position, and an alternate base. Get a likelihood ratio for each substitution, with no task-specific training.

Variant-scoring example

Extract embeddings

Provide DNA sequences. Get a representation at each nucleotide to train a probe for your task.

Embedding example

Fine-tune on your data

Use labelled sequences to adapt the model with full fine-tuning or parameter-efficient methods.

Fine-tuning scripts

Models, data, and report

Model collection Checkpoints, model cards, code examples, and licence. Pre-training corpus Sequence windows from 320 species and source assemblies. SAE model Sparse autoencoder for Botanic1-M. Technical report bioRxivTraining, evaluation, long context, and interpretability.

Cite

Barozet A., Cabeli V., Ogier du Terrail J., Rukhovich A., Janssoone T., Klajer G., Sheikhitarghi Z., Andrews G., Veran C., Strouk L. BOTANIC-1: a series of long-context plant genomic foundation models in the agentic era. bioRxiv (2026). doi: 10.64898/2026.09.04.749355

@article{Barozet2026.09.04.749355,
    author = {Barozet, Am{\'e}lie and Cabeli, Vincent and Ogier du Terrail, Jean
              and Rukhovich, Alexey and Janssoone, Thomas and Klajer, Gary
              and Sheikhitarghi, Zeinab and Andrews, Gregory and Veran, Cyril
              and Strouk, L{\'e}onard},
    title = {BOTANIC-1: a series of long-context plant genomic foundation
             models in the agentic era},
    journal = {bioRxiv},
    year = {2026},
    elocation-id = {2026.09.04.749355},
    doi = {10.64898/2026.09.04.749355},
    publisher = {Cold Spring Harbor Laboratory},
    URL = {https://www.biorxiv.org/content/early/2026/09/09/2026.09.04.749355},
    eprint = {https://www.biorxiv.org/content/early/2026/09/09/2026.09.04.749355.full.pdf},
}