Subscribing to a Built-in Commercial Service in ModelArts Studio (MaaS)

Operation Scenarios

• Enterprise intelligent customer service: Enterprises hope to optimize their customer service systems by using inference APIs to achieve intelligent Q&A and intent recognition, thereby improving customer service efficiency and customer satisfaction.
• Content creation assistance: Media and advertising companies leverage inference APIs for copywriting and creative generation, enhancing the efficiency and quality of content production.
• Intelligent data analysis: Financial and e-commerce enterprises use inference APIs to conduct in-depth analysis of massive amounts of data, uncovering data value and aiding in decision-making.

Constraints

• Subscribing to a commercial service will subscribe all versions of the service. You cannot subscribe to just one version alone.
• Currently, commercial services cannot be disabled.

Billing

Model inference services process input by converting it into identifiable tokens. Using a built-in MaaS service charges you based on token count. For details, see Model Inference Billing Items.

Coupon Instructions

• When there is a discount offer, a relevant prompt will appear on the commercial service tab of the built-in service page. When subscribing to a commercial service, available coupons will be claimed by default. During the billing process, coupons will be applied first.
• The scope of application and eligibility criteria for different coupon activities vary. Whether you can successfully claim a coupon depends on the actual activity rules.
• For information on the issuance and usage of discount coupons for model services, go to Billing Center > Promotions > Coupons.

Service Calls

The platform can send requests to different regions as needed.

Subscribing to a Commercial Service

1. Log in to the ModelArts Studio (MaaS) console and select the target region on the top navigation bar.
2. In the navigation pane on the left, choose Real-Time Inference.
3. Choose Built-in Services > Commercial Services. In the Operation column of the target service, click Subscribe.
4. In the displayed dialog box, select commercial services as required (all services are selected by default), select I have read and agree to this statement and ModelArts Studio Service Agreement, and click Subscribe.

   Selecting a commercial service will subscribe all model versions of the service.

   Table 1 Commercial service list parameters

   Parameter	Description
   Name	Name of the commercial service. Click on the left of the service name to view the service version. The name displayed in the model Parameter column can be used for model calls. Figure 1 Viewing the service version
   Subscription Status	Subscribed: The commercial service is enabled. Not Subscribed: The commercial service is not enabled.
   Type	Type of the commercial service.
   Billing Mode	Billing mode of the commercial service. The billing mode varies depending on the model. For details, see Billing Item (ModelArts Studio).
   Infer Price	Inference pricing of the commercial service. The inference pricing varies depending on the model. For details, see Billing Item (ModelArts Studio).
   Discounts	Discounts available for the commercial service. -- indicates that there is no discount.
   Model flow limiting	Total quota for accessing all services under the same model within the current account. TPM: The number of tokens processed per minute, including both input and output. RPM: The number of requests processed per minute.
   Calls	Click to go to the Service Call Details page and view the call data and monitoring metric details of the commercial service in a specified period. For more information, see Viewing the Call Data and Monitoring Metrics of Real-Time Inference on ModelArts Studio (MaaS).
   Operation	Operations supported by the commercial service. Unsubscribe: If this button is unavailable, the service cannot be disabled. No fee is generated when the service is not used. View Call Description: Click View Call Description, select a service version, and view the information and procedure for calling the commercial service. For more information, see Calling a Model Service in ModelArts Studio (MaaS).

About Arrears

When you are in arrears for using a model service, the corresponding resource instances will not immediately stop providing service; instead, the resources enter a grace period. You must pay for the pay-per-use resources used during the grace period. The related fees can be viewed in the management console > Billing Center > Overview > Arrears, and Huawei Cloud will automatically deduct the arrears amount when you top up. At this time, the resources will not be frozen, but it will affect the user's ability to subscribe to new resources or new services. Existing resources can still be used normally.

If you do not pay the arrears within the grace period, specific resources will trigger an arrears freeze and enter a retention period, with the resource status changing to Frozen. At this point, the frozen resources cannot be used, and new models that have not been subscribed to will not be supported for subscription.

After the retention period expires, if you still have not paid the account arrears, the related resources and order records under your account for this model will be cleared, and the data cannot be recovered. The paid status of the corresponding model will change to Not Subscribed.

After incurring arrears, top up your account in a timely manner. For details, see Top-Up and Payment.

Figure 2 Periods of pay-per-use resources

FAQs

1. Is there any billing example?

   For billing items and examples, see Model Inference Billing Items.

2. Can a paid service be disabled after subscription?

   Paid services cannot be disabled. No expenditures are generated for unused services.

3. What should I do if the model status is frozen when I use a commercial service?

   At this point, the frozen resources cannot be used, and new models that have not been subscribed to will not be supported for subscription. You can unfreeze the resources by recharging your account. The frozen resource instances will be restored for use, and models that have not been subscribed to will support subscription. For details, see Top-Up and Payment.

4. If the Requests Per Minute (RPM) of the model service is 300, can 300 requests be sent in one second?

   No. If the RPM of the model service is 300, it means that up to five requests can be processed per second (300/60=5). Therefore, sending 300 requests in one second would far exceed the service's processing capacity, leading to request failures.

   It is recommended to send API requests evenly to avoid sending a large number of requests in a short period. According to the API gateway's rate limiting mechanism, if the number of requests in one second exceeds RPM/60 x 1, the excess requests may trigger the API gateway's rate limiting, resulting in request failures and the return of error code 429 (Too Many Requests).