Run agents securely and efficiently with the next generation AgentCore runtime

This article announces updates to Amazon Bedrock AgentCore runtime, a managed compute layer for deploying and running AI agents securely at scale.

• Sessions start up to 15x faster with support for 8 vCPUs, 32 GB memory, and 10 GB container images
• Idle agents scale to zero with up to 68% less memory consumption per session
• Committed baseline pricing for always-on agents with burst capacity up to 4x baseline
• Data isolation extends beyond compute with FUSE-supported filesystem mounting to S3 buckets
• Each session maintains hardware-isolated Firecracker MicroVM with own identity, policy, and audit trail
• Available across North America, South America, Europe, and Asia Pacific regions

The next-generation AgentCore runtime enables enterprises to scale production AI agents with strong security isolation, efficient resource utilization, and predictable costs.