The billing for Elastic GPU Service is the same as the billing for Elastic Compute Service (ECS). This topic describes the billing details of Elastic GPU Service, including the billable items, billing methods, renewal policy, and refund policy.
Billable items and the billing methods
A GPU-accelerated instance includes computing resources such as vCPUs, memory, and a GPU, an image, and block storage devices. The following table describes the billable items of Elastic GPU Service.
Billable item | Description | Billing method | Billing method switching | Billing references |
Computing resources (vCPUs, memory, and a GPU) | You are charged for the computing resources required by an instance type, including vCPUs, memory, and a GPU. The price of an instance type varies based on the region. For more information, go to the pricing page. | Subscription, pay-as-you-go, and preemptible instance. You can use the pay-as-you-go billing method together with reserved instances or savings plans. | ||
Image | Image fees vary based on the image type and the usage duration. Images are classified into public images, custom images, shared images, and Alibaba Cloud Marketplace images. The billing for the images varies based on the image types. | Subscription and pay-as-you-go. You can use the pay-as-you-go billing method together with reserved instances. Note Images must be used only with ECS instances. The fees for Windows reserved instances include the fees for public images and can cover public image charges. | Not supported | |
Block storage device | Block storage devices are classified into cloud disks and local disks.
| Subscription, pay-as-you-go, and storage capacity unit (SCU). You can use the pay-as-you-go billing method together with savings plans. | ||
Public bandwidth | You can access the Internet by using static public IP addresses, elastic IP addresses (EIPs), and NAT gateways.
| Pay-by-bandwidth and pay-by-data-transfer. | Change the billing method for network usage of an ECS instance that uses a static public IP address | |
Snapshot |
| Pay-as-you-go, Standard locally redundant storage (LRS) storage plan, and SCU. | Not supported |
Renewal policy
You can renew only subscription instances. For pay-as-you-go instances, you do not need to renew them, but you must make sure that your account has sufficient balance. For more information, see Renewal overview.
You can enable the economical mode for pay-as-you-go instances. The fees for a GPU-accelerated instance type include the fees for GPU resources. If you enable the economical mode for the instance, you are not charged for GPU resources. For more information, see Economical mode.
The following table describes the features that are related to the renewal of subscription instances.
Feature | Scenario description | References |
Manual renewal | Before an instance is automatically released, you can renew the instance in the ECS console to extend the expiration date. | |
Auto-renewal |
| |
Renewal and configuration downgrade | If the configurations of an ECS instance exceed your requirements, you can renew the instance and downgrade the configurations within 15 days before the instance expires or the time period when the instance has expired but is not released. | |
Unified expiration date | If you have multiple subscription instances that expire on different dates, you can synchronize the expiration dates to the same day of the month to simplify management and renewal. |
Bills
You can go to Expenses and Costs to view the bills and the consumption details of your ECS resources. For more information, see View billing details.
Expiration or overdue payments
After ECS resources expire, the resource status varies and the overdue payments are handled based on the billing methods. For more information, see Overdue payments.
Billing FAQ
If billing issues occur when you purchase or use Elastic GPU Service, you can obtain information to help resolve the issues. For more information, see Billing FAQ. You can also submit a ticket to contact technical support.
For more information about the cost components, cost benefits, and cost optimization suggestions of ECS, see Best practices for cost optimization.