Amazon Web Services

#Overview

Amazon Web Services is a general-purpose cloud model hub rather than a health-curated garden, and the place to deploy, fine-tune, and serve biological foundation models at scale on AWS infrastructure. Biological models arrive through SageMaker JumpStart and Amazon Bedrock, while AWS HealthOmics orchestrates genomics and variant workflows around them. For teams already running their data and pipelines on AWS, it keeps model inference, training, and downstream analysis inside one account.

#What you can run on Amazon Web Services

ESM-3, the multimodal protein language model, is available through SageMaker JumpStart with Amazon Bedrock support, covering protein sequence, structure, and function reasoning for design and annotation tasks. Beyond dedicated bio listings, the hub hosts general-purpose foundation models, and NVIDIA BioNeMo NIM microservices for drug-discovery and protein workflows also run on AWS. HealthOmics adds managed storage and workflow orchestration for genomics data, so protein language modeling and genomics pipelines sit side by side.

#Inference and fine-tuning on Amazon Web Services

SageMaker JumpStart offers one-click deployment and fine-tuning, letting you adapt ESM-3 and other models on your own data and serve them from managed endpoints in your account. Amazon Bedrock exposes supported models as managed APIs through a serverless marketplace, and weight-bearing containers can be deployed into accounts you control. There is no dedicated biology filter, so bio models are selected from the general JumpStart and Bedrock catalogs plus partner listings, backed by AWS's enterprise security, networking, and scaling.