Announcing SOCI-indexed images for faster SageMaker AI jobs

Amazon Web Services announced Seekable OCI (SOCI) indexing for Amazon SageMaker AI Training and Amazon SageMaker AI processing jobs[1], according to a post on the company's blog.AWS wrote that "container image download is pure overhead: billable job time that produces no model progress."[1] The company added that deep learning images routinely exceed 10 GB or more[1].According to AWS, the feature fetches only the files needed to start the container and streams the rest in the background as your code runs[1]. The company said users can turn it on by pointing to a pre-indexed AWS Deep Learning Container (DLC) or adding an index to your own image[1].AWS said it tested eight images, 1.8–35.4 GB compressed[1], and claimed the indexed images had no impact on training performance[1].The company said that jobs always fall back to a standard download when needed[1], so adding an index never stops a job from running. AWS also said that existing job definitions, API calls, and data pipelines stay exactly the same[1].AWS cautioned that SOCI-indexed images run on a containerd-based runtime[1] and told users to check their scripts for code that depends on Docker.According to AWS, the feature is available across all AWS Regions[1]. In December 2025, AWS announced the same indexing for SageMaker Studio and said it cut container startup times by 30-50% when using custom images[2].