AgentsSeptember 18, 2026via AWS Machine Learning Blog
The new AgentCore runtime: Elastic, optimized, and consistently fast starts
Why it matters
Amazon Bedrock AgentCore's new runtime solves three hard problems at once: memory efficiency in long-running agent sessions, predictable latency under variable load, and cost predictability. This is agent infrastructure moving from 'can it work?' to 'can it work at scale in production.'
Key signals
- New AgentCore runtime ships with elastic memory reclamation
- Consistent cold-start latency independent of image size or concurrency
- Positioned as production-grade agent infrastructure
- Focus on speed, flexibility, and cost efficiency
- Amazon Bedrock AgentCore (AWS agent deployment platform)
- Published September 18, 2026
- Amazon Bedrock AgentCore runtime announcement
- Elastic memory reclamation on session release
- Consistent cold-start performance independent of image size or concurrency
- Focus on production-grade speed, flexibility, and cost efficiency
The hook
AWS just shipped the infrastructure layer agents need to move past pilots—elastic memory, predictable cold starts, cost control.
Today we are announcing the new AgentCore runtime, a capability of Amazon Bedrock AgentCore built for the speed, flexibility, and cost efficiency that production agents demand. It reclaims memory as sessions release it and delivers consistent cold starts regardless of image size or concurrency.