OFICIAL AWS What's New

AgentCore runtime instances are now generally available

What happened
Based on AWS What's New · Aug 06, 2026

AWS has introduced runtime instances for Amazon Bedrock AgentCore, enabling AI agents to run on custom EC2 instances with managed infrastructure, supporting sessions up to 14 days.

🚩 Report a correction ↗ Open source
Key points
·
Today, AWS announces runtime instances, a new feature in Amazon Bedrock AgentCore that lets you run agents on your own Amazon EC2 instances without managing infrastructure.
·
AgentCore runtime provides purpose-built infrastructure to deploy and operate AI agents securely at scale.
·
Using the AWS Management Console, CLI, SDKs, or APIs, you create a capacity provider that specifies the EC2 instance types your agents need, including GPU-accelerated, memory-optimized, and compute-optimized families, and attach your agents to it.
·
Runtime instances support long-running agent sessions of up to 14 days, while the default serverless, microVM-based runtime is designed for sessions of up to 8 hours that need fast startup.
Key numbers
·
Runtime instances support long-running agent sessions of up to 14 days, compared to the default serverless microVM runtime’s 8-hour limit designed for fast startup.
·
AWS has introduced runtime instances for Amazon Bedrock AgentCore, enabling AI agents to run on custom EC2 instances with managed infrastructure, supporting sessions up to 14 days.

AWS announced runtime instances for Amazon Bedrock AgentCore, allowing organizations to deploy AI agents on their own Amazon EC2 instances without handling infrastructure management. This new feature provides purpose-built infrastructure for secure, scalable agent operations, complementing the existing microVM-based runtime. Teams can now leverage the full range of EC2 instance types, including GPU-accelerated, memory-optimized, and compute-optimized options, tailored to their agent workloads. AgentCore automates provisioning, patching, scaling, and lifecycle management, reducing operational overhead while maintaining flexibility in compute selection.

Runtime instances support long-running agent sessions of up to 14 days, compared to the default serverless microVM runtime’s 8-hour limit designed for fast startup. Users can configure capacity providers via the AWS Management Console, CLI, SDKs, or APIs to specify required EC2 instance types for their agents. This enables deployment of mixed compute environments without altering agent deployment or invocation methods. The feature is available in nine AWS Regions, including US East (N. Virginia), Europe (Frankfurt), and Asia Pacific (Tokyo).

The new runtime instances are charged separately for management compute provisioning in addition to standard Amazon EC2 costs, with pricing details accessible via the AgentCore pricing page. Organizations can begin using runtime instances immediately by consulting the AWS News Blog or AgentCore documentation for implementation guidance. This expansion addresses the need for sustained, resource-intensive, or specialized-hardware agent workloads that require extended session durations.

AWS emphasizes that runtime instances integrate seamlessly with existing AgentCore workflows, allowing teams to choose the most suitable compute environment for each agent without additional deployment complexity. The feature aims to enhance operational efficiency by offloading infrastructure management to AWS while providing granular control over compute resources.

Original source → Deals on Clipraptor.com →