← All news
Global · AI & Technology
Global·AI & Technology

AWS updates AgentCore runtime with faster starts and more compute

AWS announced updates to Amazon Bedrock AgentCore runtime that increase session capacity and reduce startup times. The company says sessions retain hardware isolation while supporting larger agent workloads.

AI Hubs United Editorial ·

AWS announced updates to Amazon Bedrock AgentCore runtime, the managed compute service for deploying and running AI agents. According to the AWS Machine Learning Blog, the updated runtime starts sessions up to 15 times faster and supports four times as much compute as the previous version. Each session continues to run in a separate hardware-isolated Firecracker micro virtual machine, with its own identity, policy and audit trail.

Sessions now support up to 8 vCPUs and 32 GB of memory, alongside container images as large as 10 GB. AWS says startup takes less than two seconds regardless of image size or concurrency. For teams running code compilation, browser-based workflows or local model inference, these changes provide more capacity within an isolated session and could reduce the need to maintain pre-provisioned infrastructure to avoid startup delays.

AWS also reported customer results. Mission Cloud said an initial comparison using the same natural-language query workload showed 35% faster cold starts for inference requests and a 17% reduction in measured runtime cost per call. These results offer a workload-specific example of the potential operational benefits, rather than a universal performance guarantee.

AI Hubs United original short summary of the source linked above. It is not the original article; read the full story at the source.