Find your GCP compute option

Compare Google Cloud compute options by abstraction level, billing model, stateful workload fit, runtime control, and release workflow.

Find your GCP compute option

Compare Google Cloud compute options by abstraction level, billing model, stateful workload fit, runtime control, and release workflow.

Showing 8 of 8 vendors that match

1st

Compute Engine (GCE VMs)

Infrastructure-as-a-service virtual machines for maximum OS, networking, storage, and lifecycle control.

Free

Always Free includes limited e2-micro usage in eligible regions; production VMs are pay-as-you-go by machine, disk, and network.

Best fit when you need full VM and OS control
Supports stateful services, custom images, GPUs, and specialized networking
Broad machine family and committed-use options

You own patching, autoscaling design, and runtime operations
Always-on resources can cost more for spiky web workloads
Canary routing usually needs external load balancing or deployment tooling

Visit

2nd

GKE Autopilot

A more managed GKE mode where Google manages nodes and charges closer to requested workload resources.

$0/mo

Usage-based pricing for Autopilot workload resources; minimum monthly cost depends on deployed pods and regions.

Kubernetes API with less node management
Good middle ground for teams that need Kubernetes but not node tuning
Supports many production microservice patterns

Less infrastructure control than GKE Standard
Autopilot security and configuration constraints can reject some workloads
Still more complex than serverless options for basic web apps

Visit

3rd

Cloud Run

A fully managed container platform for stateless services, jobs, and request-driven workloads.

Free

Generous free tier; request-based billing can scale to zero, while instance-based billing charges for instance lifetime.

Simple deployment for any HTTP container
Scales to zero and supports revision traffic splitting
Strong default for stateless APIs and web apps

Local filesystem is disposable and not suitable for durable state
Cold starts and request limits need design attention
Less control over host-level networking and daemons than VMs or GKE

Visit

4th

Cloud Run Functions

Source-deployed functions that run as Cloud Run services for HTTP and event-driven handlers.

Free

Billed on Cloud Run pricing for generated services, with free-tier usage applying where eligible.

Fast path for small HTTP or event functions without container authoring
Built on Cloud Run services and Google buildpacks
Good for glue code, webhooks, and event handlers

Less runtime and container control than Cloud Run services
Function shape can be awkward for larger applications
Still needs external storage for durable state

Visit

5th

App Engine Standard

A high-level platform-as-a-service for supported language runtimes with automatic scaling and version traffic splitting.

Free

Standard environment has free quotas; paid usage starts after free tier and depends on instance class and traffic.

Very managed deployment model for supported runtimes
Built-in versions and traffic splitting
Can be inexpensive for small always-available apps within quotas

Runtime and platform constraints are tighter than Cloud Run
Less portable than standard containers
Not a good fit for stateful or long-running background services

Visit

6th

Cloud Functions (Gen 2)

The Cloud Functions v2 API path for event-driven and HTTP functions backed by Cloud Run infrastructure.

Free

Gen 2 functions use Cloud Run-style billing; exact cost depends on invocations, CPU, memory, duration, and networking.

Natural fit for Google Cloud event triggers
Familiar Cloud Functions API and deployment model
Backed by newer Cloud Run infrastructure

Cloud Run functions are the newer recommended surface for many new function deployments
Less flexible than deploying a container directly to Cloud Run
Traffic splitting is not the main workflow through the Cloud Functions API

Visit

7th

App Engine Flexible

A VM-backed App Engine environment for custom runtimes with more control than Standard but less than raw Compute Engine.

$35/mo

No App Engine Flexible free tier; VM resources are billed per second with a one-minute minimum.

Supports custom runtimes with App Engine deployment conventions
Traffic splitting and versioning remain built in
Useful for legacy App Engine apps needing more runtime flexibility

No free tier and typically higher idle cost than Cloud Run
VM-backed scaling is slower than Standard or Cloud Run
Less direct infrastructure control than Compute Engine

Visit

8th

Google Kubernetes Engine (GKE Standard)

Managed Kubernetes control plane with direct node pool control for teams that need Kubernetes flexibility.

$73/mo

Cluster management is commonly about $0.10/hour after free allowances; nodes, storage, and networking are billed separately.

Full Kubernetes API with managed control plane
Strong fit for complex microservices and platform teams
Supports stateful sets, custom nodes, service mesh, and progressive delivery

Requires Kubernetes operations expertise
Node-based costs remain even when traffic is low
More moving parts than Cloud Run for simple services

Visit

About this comparison

Compare Compute Engine, GKE Standard, GKE Autopilot, Cloud Run, Cloud Run functions, App Engine Standard, App Engine Flexible, and Cloud Functions Gen 2. Filter by budget, team size, abstraction level, billing model, stateful workload support, scaling latency, custom runtime support, and traffic splitting to choose the right Google Cloud compute platform.