Yearly/Monthly

Application Scenario

The yearly/monthly billing mode requires you to pay for a subscription period before using the resources. It is suitable for long-term users with stable demands. The following are some service scenarios applicable to this billing mode:

• Stable resource demands: For services with long-term, stable resource requirements, such as large model training that requires a long time, the yearly/monthly billing mode offers high cost-effectiveness.
• Long-term projects: For long-term projects, such as scientific model training, the yearly/monthly billing mode ensures stable resource usage throughout the project period.

Constraints

Yearly/Monthly resource pools cannot be used across regions.

Applicable Billing Item

Compute resources (vCPUs and GPUs)

ModelArts provides yearly/monthly dedicated resource pools. To purchase one, do as follows: On the ModelArts management console, choose Dedicated Resource Pools > Elastic Cluster. On the Resource Pools tab, click Create. Then, set Billing Mode to Yearly/Monthly and view the price details in the lower left corner of the page. The price is calculated based on the resources selected for the dedicated resource pool, including vCPUs and GPUs.

Billing Cycle

The billing cycle of yearly/monthly resources is determined by the purchase duration (UTC+08:00). A billing cycle starts from the time when you enable or renew a resource (accurate to seconds) and ends at 23:59:59 on the expiration date.

For example, if you purchase a dedicated resource pool for one month at 15:50:04 on March 8, 2023, the billing cycle is from 15:50:04 on March 8, 2023 to 23:59:59 on April 8, 2023.

Billing Example

Assume that you purchase a yearly/monthly dedicated resource pool on March 8, 2023 at 15:50:04. The flavor GPU: 1*nvidia-t4 | CPU: 8 vCPUs 32GB, and one compute node are used. The billing resource is GPU. Your subscription is one month and you manually renew it for another month before it expires. The billing cycles are as follows:

• The first billing cycle is from 15:50:04 on March 8, 2023 to 23:59:59 on April 8, 2023.
• The second billing cycle is from 23:59:59 on April 8, 2023 to 23:59:59 on May 8, 2023.

You must prepay for each billing cycle. Table 2 shows the billing formula.

The fee in the preceding example is calculated as follows:

Fee of the yearly/monthly dedicated resource pool = 625.10 x 1 x 2 = $1250.20 USD

The prices in this document are for reference only. The actual prices are subject to ModelArts Pricing Details.

Billing Impacts Following Configuration Changes

If the specifications of a yearly/monthly compute resource do not meet your service requirements, you can change them on the ModelArts management console and pay for the new specifications.

• Upgrades: You change the current specification of a resource to a new, more expensive specification and then pay for the difference.
• Downgrades: You change the current specification of a resource to a new, less expensive specification and Huawei Cloud refunds the difference.

Avoid degrading resource specifications as it will negatively affect ModelArts performance. Example of upgrade fee (without any discount): Assume that you purchase a dedicated resource pool for one month on April 8, 2023. The flavor GPU: 1*nvidia-t4 | CPU: 8 vCPUs 32GB, and one compute node are used. On April 18, 2023, you add one compute node. The price of the configuration becomes $1,250.20 USD/month from $625.10 USD/month. The formula is as follows:

Upgrade fee = Price of the new specification x Remaining duration – Price of the old specification x Remaining duration

The remaining duration in the formula is the sum of the remaining days in each calendar month divided by the maximum days in the corresponding calendar month. In this example, the remaining duration = 12 (Remaining days in April)/30 (Maximum days in April) + 8 (Remaining days in May)/31 (Maximum days in May) = 0.6581. The specification upgrade fee is calculated as follows: 1,250.20 x 0.6581 – 625.10 x 0.6581 = $411.37 USD.

For details, see Pricing of a Changed Specification.

Impacts of Expiration

Figure 1 describes the status of yearly/monthly resources in each period. After purchasing a resource, it remains operational throughout the billing cycle. This can be referred to as the validity period. If the resource is not renewed after expiration, it enters the grace period and then the retention period.

Figure 1 Periods of yearly/monthly resources

Expiration warning

You will be notified seven days before your yearly/monthly dedicated resource pool expires. Warning messages will be sent by email, SMS, and internal message.

Impacts of expiration

If your yearly/monthly resource pool is not renewed after expiration, it first enters the grace period and its status changes to Expired. During the grace period, you can access ModelArts, but the following operations will be restricted:

• Changing resource pool specifications
• Unsubscribing from the resource pool
• Changing the billing mode of the resource pool

If you do not renew the yearly/monthly resource pool within the grace period, the resource enters the retention period and the resource status changes to Frozen. You cannot perform any operation on the yearly/monthly resource in the retention period.

If the yearly/monthly resource pool is not renewed after the retention period ends, compute resources (vCPUs and GPUs) will be released and data cannot be restored.

• Both the grace period and retention period are 15 days.