Cloud Run vs App Engine Cost: Choose the Right Runtime

Cloud Run can use request-based billing, which charges allocated CPU and memory while an instance starts, stops, or handles requests, plus a request charge. With no minimum instances, an idle service can scale to zero. Instance-based billing instead charges for the full container instance lifecycle and can fit steady traffic or work that needs CPU outside requests.

App Engine Standard charges instance hours by instance class after its applicable free usage, while App Engine Flexible bills its virtual machine resources and has no free tier. That means a low-traffic comparison should include idle instance time and scaling settings, not only requests per month.

Cloud Run request-based billing is usually the clearer starting point for containerized APIs with sporadic or spiky traffic. Concurrency lets several requests share one instance, while maximum instances can limit how quickly compute and downstream database connections expand. Minimum instances reduce cold starts but create an idle cost that belongs in the estimate.

App Engine Standard can suit applications that fit its supported runtimes and platform conventions and benefit from instance-class autoscaling. App Engine Flexible offers more runtime flexibility, but its VM-based resources make continuously running capacity important to model. Do not choose between them from a single per-request price: include response time, memory, instance count, and hours active.

For both products, add outbound data transfer, Cloud Logging, builds and deployments, and the database or storage layer. A runtime that looks inexpensive can still produce a larger bill through Cloud SQL connections, Firestore operations, Cloud Storage downloads, or Serverless VPC Access.

Create a low-traffic and a steady-traffic scenario, then compare them in the same region. For Cloud Run, record the billing mode, CPU, memory, concurrency, request duration, and minimum and maximum instances. For App Engine, record Standard or Flexible, instance class or VM resources, scaling mode, and expected instance hours. Confirm the final choice in the official Google Cloud pricing calculator because rates and free usage vary by product and region.