AWS First Steps
LanguageEnglish
New feature

New AgentCore Runtime Now Available in Amazon Bedrock AgentCore

Amazon Bedrock AgentCore now offers a next-generation serverless microVM compute with improved memory management and faster cold starts.

Today, AWS launched a next-generation AgentCore runtime, a serverless micro VM compute within Amazon Bedrock AgentCore. This new runtime reuses unused memory during sessions, enabling billing only for actual usage and providing consistent cold start times independent of container image size or parallelism. It maintains the existing serverless model while reducing costs and shortening startup times. Each session starts with a small, efficient memory profile, allocating additional memory as needed. For cold starts, the agent environment is prepared once and a snapshot is saved, allowing new instances to restore from this snapshot instead of repeating the full startup sequence, ensuring consistent start times regardless of image size. Testing showed the new runtime achieved a P75 cold start time of 1.9 to 2.0 seconds for container images ranging from 200 MB to 2 GB, compared to 5.4 to 30 seconds in V1. The new AgentCore runtime is available in us-east-1, us-east-2, us-west-2, eu-west-1, and ap-northeast-1 regions. Users can start creating or updating the runtime by setting the platform version to V2.

Why it matters

Amazon Bedrock AgentCore is an AWS service for building AI agents. This update improves performance and cost efficiency, enabling developers to build agents more efficiently.

Read the original AWS announcement

Related articles and terms

Understanding Lambda Cold Starts and Choosing Among Three Optimization StrategiesThis article explains the mechanism behind Lambda cold starts from the Firecracker MicroVM lifecycle perspective, and compares optimization techniques across three axes: SnapStart, Provisioned Concurrency, and function design in terms of cost and constraints.AWS LambdaA serverless compute service that runs code without server management, using a pay-per-use model based on request count and execution durationAmazon BedrockA fully managed generative AI service providing access to leading foundation models via APIBehind Lambda's 1,000 Concurrent Executions - How the Firecracker Warm Pool and the Worker Manager WorkThis article explains how Lambda manages microVMs within the default limit of 1,000 concurrent executions, the reuse strategy through the warm pool, the placement decisions of the worker manager, and the internal optimizations that lower the cold start rate.Getting Started with Serverless Development Using AWS Lambda - Function Design and Event Source IntegrationThis article covers Lambda function design, event source mapping, cold start mitigation, and Powertools usage.