AWS / update detail

Amazon SageMaker AI announces training session APIs for serverless model customization

Official AWS changelog summary with source attribution, release tags and upgrade verdicts.

Track this AWS release note alongside related changelog updates, risky changes, and weekly digest signals from the developer community.

AWS·Oct 22, 2026, 04:30 PMFeatureAI-assisted summary

Summary

Introduces serverless training session APIs enabling custom training loops (e.g., SFT, DPO, RL) with API calls for forward/backward, optimizer steps, and sampling while SageMaker manages GPU allocation and fault recovery.
Adds per‑token billing for processed tokens and releases a training sessions cookbook with ready‑to‑run SFT and RL recipes.
Makes the feature generally available in US East (N. Virginia), US West (Oregon) and Asia Pacific (Mumbai), accessible via the SageMaker Python SDK or Studio.

featuredocssource healthy

Official source Last checked Oct 23, 2026, 3:01 AM

Loading follow options…

keep exploring

Continue with AWS

Compare this release with the product timeline and keep the tools you use in one place.

official source

Every summary should remain traceable to the original changelog or release source.

AI-assisted summarysource healthyLast checked Oct 23, 2026, 3:01 AM

digest settings

Already following AWS? Add related tools without starting a new digest.