Amazon SageMaker JumpStart

Amazon SageMaker JumpStart is a model and solution hub within Amazon SageMaker AI for discovering, evaluating, customizing, and deploying pretrained foundation models, task-specific models, and built-in algorithms.

AI Practitioner focus

  • Reduces time to start from an existing model instead of training from scratch.
  • Includes open-weight/public models, third-party proprietary models, and classic ML algorithms; licenses and subscription requirements vary.
  • Models deploy to SageMaker AI endpoints, so the customer chooses and pays for the hosting infrastructure and controls the broader ML lifecycle.
  • Supports selected fine-tuning/customization and integration with SageMaker evaluation and MLOps capabilities.
  • JumpStart vs Amazon Bedrock: JumpStart emphasizes model deployment/customization inside SageMaker-managed ML infrastructure; Bedrock provides serverless API access to supported FMs and managed GenAI building blocks with less infrastructure management.

Sources