Cloud Run Instances: A Persistent Runtime for AI Agents

Google Cloud has launched Cloud Run instances, a compute runtime built for long-lived, stateful AI agents that need to stay awake without paying for a full 24/7 VM.

Agents like OpenClaw or Hermes are built for one user and need to run continuously — they don’t fit the scale-to-zero model that works well for stateless APIs. Cloud Run instances run as a single, persistent instance instead of autoscaling, which matches what a personal agent needs better than either a stateless Cloud Run service or a dedicated VM.

What Cloud Run Instances Offer

• Singleton mode: exactly one instance, no autoscaling.
• Up to 7-day continuous runtime, with automatic restarts.
• A stable HTTPS URL that survives updates and restarts.
• Stop and resume on demand, so you can pause an agent when it isn’t needed.

Pricing starts at $5.70 a month for 1 vCPU and 1 GiB of memory running 24/7. The shared vCPU uses burst budgets, so an idle agent costs little and can still spike in compute when it’s doing real work.

Deploying OpenClaw

OpenClaw is an open-source personal AI agent that improves with use. Run locally, it stops the moment you close your laptop; Cloud Run instances give it a persistent home instead.

After uploading your config to a Cloud Storage bucket, deploy it with one gcloud command:

gcloud beta run instances create openclaw-instance
  --image ghcr.io/openclaw/openclaw:latest
  --port 18789
  --public
  --add-volume mount-path=/home/node/.openclaw,type=cloud-storage,mount-options="uid=1000;gid=1000;file-mode=0700;dir-mode=0700",bucket=${BUCKET}
  --set-env-vars "OPENCLAW_GATEWAY_PASSWORD=${PASSWORD},GEMINI_API_KEY=${GEMINI_API_KEY}"

Once deployed, the agent can be reached over Telegram, WhatsApp or any platform you connect it to. Full instructions are in this codelab.

Early Usage

OffDeal, an AI-powered investment bank for small businesses, is already using Cloud Run instances for its long-running agents, reporting an 88% reduction in cold starts and describing the setup as “very straightforward and reliable.”

What’s Next

SSH access for both instances and services is coming; sign up for private access for early entry. Cloud Run instances are currently in preview.

See the official docs for details, or our Google Cloud review for how this fits into a broader GCP stack.

Original announcement published on Google Cloud.