Google Cloud Agent Sandbox

https://cloud.google.com/

GKE Agent Sandbox manages isolated, stateful, single-replica workloads on Google Kubernetes Engine. Developers use it to run AI-agent code, development environments, and interactive tools in sandboxes.

Pricing model Per-second charges for requested resources of running Pods, plus GKE cluster fees
Pricing The headline compute price as the provider states it. Sorting ranks prices stated per vCPU over time, converted to hourly. Other pricing models sort after, unranked. GKE Autopilot general-purpose, regular price: $0.0445 per vCPU-hour
Free while idle Whether a stopped, paused, or sleeping sandbox costs nothing. ✕
Elastic Resources flex with what the sandbox actually uses, and the bill follows: yes when the provider bills on active or observed use, no when capacity is reserved or allocated and billed while it runs. ✕
Memory snapshots Whether RAM state survives a pause and resume, not just disk. ✓
Wake on request Whether a stopped sandbox wakes automatically on inbound traffic. ✓
Start / resume Typical time from create or resume to running. Sorting ranks stated times, converted to seconds; claims with no figure sort after, unranked. ~200 ms for 90% of warm-pool allocations; under 1 second with pre-warmed Pods
Max runtime The longest a sandbox may run. No fixed limit; runs until deleted unless a shutdown time is set
Isolation The isolation technology between sandboxes. Runtime-configurable kernel-level isolation, typically gVisor; Kata Containers is also supported
GPUs Whether GPU instances are available for sandboxes. ✓
Docker Whether Docker containers can run inside the sandbox. ✓
SDKs Python, Go

✓ yes · ✕ no · – no cited public fact.

Full feature support and snapshot compatibility depend on GKE version and configuration; Pod snapshots have machine-type, GPU, and storage limitations, and GKE stores snapshot data in Cloud Storage.

Sources

Last verified 2026-10-02. Corrections land through the update log.