OFICIAL Google Cloud Blog AI & Software · Aug 27, 2026

Introducing Cloud Run instances

What happened

Based on Google Cloud Blog · Aug 27, 2026

Google Cloud introduced Cloud Run instances to support long-lived, stateful workloads like AI agents, offering a lower-cost alternative to dedicated VMs with predictable pricing and continuous HTTPS endpoints.

🚩 Report a correction ↗ Open source

Google Cloud Blog — Google

Key points

Need a low-cost, high-performance way to run long-lived, stateful workloads such as AI agents?

Today, we introduced Cloud Run instances, which let you do just that.

Consider AI agents such as OpenClaw or Hermes, which are intended for individual developers or personal use.

Because these agents often work continuously and tend to serve only one user at a time, their infrastructure requirements look quite different from stateless, high-throughput web services that typically run on Cloud Run services.

Key numbers

The pricing for a Cloud Run instance with 1 vCPU and 1 GiB of memory running continuously for 30 days is set at $5.

Early adopters, such as OffDeal’s Luis Ruiz Morel, have reported significant improvements, including an 88% reduction in cold starts and straightforward implementation.

Google Cloud has launched Cloud Run instances, a new offering designed to run long-lived, stateful workloads such as AI agents like OpenClaw or Hermes. Unlike Cloud Run services, which scale to zero when idle, instances maintain a single, continuously running copy, making them suitable for agents that require persistent availability. The pricing for a Cloud Run instance with 1 vCPU and 1 GiB of memory running continuously for 30 days is set at $5.70, utilizing shared vCPU with burst budgets for cost efficiency.

The new instances provide a fixed HTTPS URL that remains unchanged across updates and restarts, eliminating the need for manual provisioning of endpoints. This model is particularly beneficial for agents that operate intermittently, such as personal AI assistants, which only require compute resources during active tasks. Users can deploy agents like OpenClaw with minimal effort, uploading configuration files to Cloud Storage and executing a single command to initiate the instance.

Cloud Run instances support interaction via platforms like Telegram, WhatsApp, or other social media channels, and can be connected to additional tools as needed. Google Cloud has also announced upcoming SSH access for both instances and services, currently available in private preview. The feature aims to enhance usability for developers managing long-running workloads without sacrificing performance or reliability.

Early adopters, such as OffDeal’s Luis Ruiz Morel, have reported significant improvements, including an 88% reduction in cold starts and straightforward implementation. Cloud Run instances are currently in preview, offering a cost-effective solution for running stateful workloads while maintaining high performance and predictable costs.

Original source → Deals on Clipraptor.com →