Billing

ModelArts is a one-stop development platform for AI developers. With data preprocessing, semi-automated data labeling, distributed training, automated model building, and model deployment on the device, edge, and cloud, ModelArts helps AI developers build models quickly and manage the lifecycle of AI development.

ModelArts supports two billing modes: pay-per-use (billed based on the actual duration) and yearly/monthly payment (more economical).

Billing Items

ModelArts bills you for your selected resources. Table 1 lists billing items. For details about the price of each billing item, see Product Pricing Details.

Table 1 Billing items

Billing Item

Description

AI development lifecycle

Dedicated for developers with AI development experience. It supports machine learning and deep learning algorithm development and deployment, including data processing, model development, training, and management, and service deployment. The billing items include model development environments (Notebook), model training (training and TensorBoard jobs), and service deployment (real-time services).

ExeML

Dedicated for developers with few AI development capabilities. It supports automatic design, optimization, and training of models, as well as service deployment, delivering code-free AI development. This billing item applies only to training and deployment of ExeML jobs. The billing items include ExeML projects (training jobs and service deployment).

Currently, only pay-per-use billing mode is supported.

Billing Modes

ModelArts can be billed by the following modes:

  • Pay-per-use: allows for flexibility. In this mode, you can enable or disable ModelArts as needed.
  • Prepaid resource package: A subscriber purchases a certain resource quota in advance. During on-demand use, the system deducts the quota first. Resources used in excess will be billed on a pay-per-use basis.
  • Yearly/monthly: billed on a yearly or monthly basis. This mode provides a larger discount than pay-per-use.

    Currently, only the dedicated resource pool supports the yearly/monthly billing mode.

Changing the Billing Mode

When using ModelArts, you can select appropriate resources as required. ModelArts provides the following methods for you to change bill modes after a job is started:

  • If the resources you have purchased cannot meet service requirements, purchase resources with higher specifications. See Buying a Package.
  • A dedicated resource pool in Yearly/Monthly billing mode does not support scaling. If you purchase a pay-per-use dedicated resource pool, you can manually scale in or out the dedicated resource pool. You are billed based on the number of new nodes. For details, see Scaling a Dedicated Resource Pool.

If the configuration change methods provided by ModelArts do not meet your requirements, you can create a job again and migrate data to realize configuration change.

Renewal

ModelArts provides pay-per-use and pre-paid resource package subscription. In the pay-per-use mode, fees are deducted every hour and an insufficient balance may cause overdue payments. For a pre-paid resource package, when you use up the quota of this package, the system will automatically bill you in pay-per-use mode. The service will not be interrupted as long as your account balance is sufficient. If your subscription is not renewed, your services will keep running, but enter into a retention period, during which ModelArts will stop running but data is retained.

Expiration and Overdue Payment

  • Resources with the pay-per-use and prepaid resource package billing mode will not expire. If resources in a prepaid resource package have been used up, subsequent usage of the resources will be billed on a pay-per-use mode. In the pay-per-use mode, fees are deducted every hour. If your account balance is insufficient to pay for the expense occurred in the last hour, your account will be in arrears, and ModelArts has a retention period. If resources are renewed within the retention period, they will be available and you are bill from the original expiration date.

If your account is in arrears, some operations will be restricted. You are advised to renew your account as soon as possible. Table 2 describes the restricted operations.

Table 2 Restricted operations due to arrears

Function

Restricted Operation

ExeML

Model training and deployment

Data Management > Datasets

One-click model deployment

DevEnviron > Notebooks

Creating and starting notebook instances

Training Management > Training Jobs

Creating training jobs

Training Management > Auto Search Jobs

Creating auto search jobs

Service Deployment > Real-time Services

Deploying real-time services

Dedicated Resource Pools

Creating dedicated resource pools

AI Market > ModelHub

Creating models, algorithms, or HiLens skills