Bunker Longevity: Frontier AI Models for Life Sciences - Bunker Longevity

Bunker Longevity builds foundation models for the life sciences. Every model ships behind a single hosted API and a typed Python SDK. Predict a protein structure, call variants, generate a molecule, or reconstruct a cryo-EM map with the same three calls: deploy, predict, status.

Install

pip install bunker-fold

Predict

from bunker_fold import Bunker

client = Bunker()

result = client.models.bunkerfold_3.predict(
    sequence="MVLSPADKTNVKAAWGKVGAHAGEYGAEALERMFLSFPTTKTYFPHFDLSH"
)

print(result.confidence)      # 0.97
print(result.fold_class)      # 'globin'
print(result.structure_url)   # https://cdn.bunkerbio.com/predictions/...pdb

Model catalog

Model	Domain	Parameters	Endpoint
BunkerFold-3	Protein structure	1.2B	`bunkerfold-3`
GenoScan-LLM	Variant calling	800M	`genoscan-llm`
MolGen-7	Generative chemistry	3.4B	`molgen-7`
CryoNet	Cryo-EM reconstruction	540M	`cryonet`
ToxPredict-v2	Toxicity screening	280M	`toxpredict-v2`
EpiMark	Epigenomic annotation	420M	`epimark`

Base URL

https://api.bunkerbio.com/v1

All requests use HTTPS bearer token auth. See Authentication to create a key.

Endpoints

POST  /v1/models/{model}/deploy    # start an inference instance
POST  /v1/models/{model}/predict   # run a prediction
GET   /v1/models/{model}/status    # health and queue depth
POST  /v1/models/{model}/stop      # release resources

Contact

Reach the team at [email protected].