Cloud Run

Fully managed serverless platform for running stateless containers that scale automatically from zero to handle web requests, gRPC, Pub/Sub events, and batch workloads. Deploy from source or container images in Go, Node.js, Python, Java, .NET, or Ruby and pay only for the compute time your code actually runs.

Trusted by enterprises such as Telegraph Media Group, which reports substantial cost and carbon reductions after adopting serverless containers. Cloud Run sits between Cloud Functions and GKE: it offers container-based workloads with more control than functions, and simpler operations than full Kubernetes. Google Cloud Functions now runs on Cloud Run under the hood, giving function workloads access to the same service configuration options.

Key capabilities:

• Scale from zero with request-based or event-driven invocation via HTTP, gRPC, WebSockets, and Pub/Sub
• Source-based deployment with buildpacks, no Docker required for supported runtimes
• Cloud Run Jobs for run-to-completion batch processing with parallel instances and jobs up to 24 hours
• On-demand NVIDIA L4 GPUs for AI inference workloads such as LLMs and image recognition
• Direct VPC connectivity for private access to databases and internal services
• Multi-region deployment for high availability

Engineers use Cloud Run for web applications and APIs, microservices, event-driven data pipelines, scheduled or ad-hoc batch jobs, and AI inference. Integrates with Cloud Build for CI/CD, Artifact Registry for images, and Eventarc for event routing. Suits teams that want container flexibility without managing clusters.