SageMaker Async Inference now accepts inline payloads

Amazon SageMaker AI Async Inference now supports inline request payloads, letting customers send input data directly in the InvokeEndpointAsync request body.

Previously, teams often had to upload input data to S3 before each asynchronous invocation. Removing that step can simplify application code and reduce friction for workloads where payload sizes fit the new path.

For ML platform teams, small integration changes like this can matter because inference reliability often depends on reducing the number of moving pieces around the model endpoint.