Amazon SageMaker AI introduced instance preference lists for training and processing jobs, allowing users to provide multiple acceptable instance types ordered by priority. This approach helps workload jobs start more quickly by launching on the first available preferred instance, addressing capacity constraints and high demand periods effectively.
Previously, users had to request one specific instance type and implement custom retry logic to find available capacity. Now, SageMaker can also source capacity from on-demand or reserved Flexible Training Plans transparently. This feature integrates seamlessly through existing APIs, CLI, SDKs, and the console, available across all SageMaker-supported AWS Regions.
Source & further reading
Amazon SageMaker AI now supports instance preference lists for training and processing jobs ↗Source published 2026-09-15. This report was prepared with AI assistance from the linked primary-source material. Claims and limitations should be checked against the original announcement or research.