App Engine pricing depends on the environment you choose. Standard charges per instance-hour, by instance class, after a daily free quota. Flexible charges for the vCPU, memory, and disk its VMs use, with no free quota. Compute Engine charges per second for virtual machines, and AWS charges for the EC2 instances and related resources behind Elastic Beanstalk. None of these is cheapest in every case. The answer depends mostly on how steady your traffic is.
How App Engine Standard is billed
Standard bills instance-hours by instance class. These are approximate on-demand rates in US regions:
| Instance class | Approximate price per hour |
|---|---|
| F1 or B1 | about $0.05 |
| F2 or B2 | about $0.10 |
| F4 or B4 | about $0.20 |
| F4_1G or B4_1G | about $0.30 |
| B8 | about $0.40 |
Rates vary by region and change over time, so check the current pricing page or the calculator before you budget. Incoming traffic is free. Outbound traffic beyond the free allowance is billed per gigabyte.
Idle instances can keep running for up to 15 minutes after the last request, and that time is billed. Standard instances do not receive sustained-use or committed-use discounts.
The free quota
Standard includes 28 F-instance hours, 9 B-instance hours, and 1 GB of outbound traffic each day. That covers a small app that scales down when idle. More instances or a busier app will use the quota up quickly.
Caps and warm instances
Two settings in app.yaml control spend:
max_instancescaps how many instances can run, which caps your cost.min_instanceskeeps instances warm so the first request after a quiet period responds faster. Those instances cost money even when no one visits.
How Flexible is billed
Flexible bills per second for vCPUs, memory, and persistent disk, with a one-minute minimum. Approximate US rates are about $0.053 per vCPU-hour and about $0.007 per GB-hour of memory. At least one instance always runs, and there is no free tier, so a Flexible app costs money around the clock.
How Compute Engine is billed
Compute Engine bills each VM per second, with a one-minute minimum. A few details matter when you compare it with App Engine:
- Always Free: one e2-micro per month in us-west1, us-central1, or us-east1, plus 30 GB of standard persistent disk.
- Rough on-demand prices: an e2-micro costs roughly $6 to $8 a month, and an e2-medium roughly $24 to $25 a month, in typical US regions.
- Discounts: sustained-use discounts apply automatically to some machine types. Committed-use discounts require a one- or three-year commitment. Spot VMs cost much less, but Google can stop them at any time.
- Stopped VMs still cost money for their disks and static IP addresses.
The Compute Engine explainer covers machine types and storage in more detail.
How AWS is billed for comparable setups
EC2 instances are billed while they run, under on-demand, Reserved Instance, Savings Plan, or Spot terms. Storage, load balancers, and data transfer are billed separately. Elastic Beanstalk has no fee of its own. You pay for the EC2, load balancer, and EBS resources it creates.
Outbound data transfer to the internet costs about $0.09 per GB in many regions after the first 100 GB each month. Google Cloud’s outbound rates are in a similar range, roughly $0.08 to $0.12 per GB depending on volume and region. Egress is one of the costs to check on any platform.
Side-by-side
| App Engine Standard | App Engine Flexible | Compute Engine | AWS EC2 with Elastic Beanstalk | |
|---|---|---|---|---|
| Billing unit | Instance-hour by class | vCPU, memory, and disk per second | VM per second, 1-minute minimum | Instance-hours plus related resources |
| Free allowance | Daily F and B hours and 1 GB outbound | None | One e2-micro per month in three US regions, plus 30 GB disk | Varies by account and offer |
| Scales to zero | Yes | No | No | No; an environment keeps at least one instance |
Which option is cheaper?
- Spiky or low traffic: Standard often costs the least, because it can scale to zero and includes a free quota.
- Steady, always-on load: a VM often costs less, because sustained-use and committed-use discounts apply to VMs, and Standard gets neither.
- AWS comparisons: list prices for similar instance sizes are close to Compute Engine’s. The difference in a real bill usually comes from commitments, data transfer, and how the architecture is built, not from list prices alone.
The most reliable answer comes from the Google Cloud pricing calculator or AWS’s calculator, using your own traffic numbers.
Hidden costs and budgets
Instance-hours are not the whole bill. Databases, storage, outbound traffic, Cloud Build during deployments, and logs are billed separately. Budgets send alerts when spending crosses a threshold, but they do not stop spending. Check billing reports regularly.
Is Google App Engine free?
Partly. Standard has a daily free quota that can cover a small app that scales down when idle. Flexible has no free tier. New Google Cloud customers also get a $300 credit valid for 90 days. The cost guide explains the trial and the Always Free limits.
How much does Google charge for cloud services?
There is no single price. Google Cloud charges for each resource you use, such as instance-hours, vCPU and memory time, storage, data transfer, and requests. The price depends on the service and the configuration you pick.
Next steps
If you are deciding between App Engine and Cloud Run for a new project, the App Engine vs Cloud Run guide covers how the two bill differently. The Standard vs Flexible comparison explains which environment fits your traffic. If you are comparing with AWS, the Elastic Beanstalk comparison explains what each service keeps running. For the AWS side of the bill, read why AWS costs can climb.