Help Center/ PanguLargeModels/ User Guide/ Preparations/ Subscribing to the PanguLM Service
Updated on 2025-07-28 GMT+08:00

Subscribing to the PanguLM Service

Description

ModelArts Studio allows you to subscribe to model assets, data resources, training resources, and inference resources on a yearly/monthly basis. You can also subscribe to resources on a yearly/monthly or pay-per-use basis.

  • Model assets: Model assets can be used in the model development and application development modules. Currently, NLP models can be subscribed to.
  • Data resources: Data general computing units are used for regular expression operator processing during data processing. Data intelligent computing units are used for AI operator processing during data processing. Data hosting units are used for storing datasets during data engineering.
  • Training resources: Training units can be used for model training and compression of all large models.
  • Inference resources: Inference units can be used for model inference of NLP and CV models. Model instances can be used for model inference of scientific computing models.

Subscription Scenario

ModelArts Studio provides free models and paid models. After subscribing to the models and resources, you can use the data engineering, model development, and agent development functions on the platform.

You can subscribe to products on demand based on application scenarios.

Table 1 Subscription scenarios

No.

Scenario

Description

1

Data scenario

Only data resources are subscribed to. Models are not subscribed to. Scenarios such as data processing, synthesis, labeling, and combination are supported.

2

Inference scenario

In the pure inference scenario, you can use free models or subscribe to model assets (optional) and inference resources for inference after deployment.

3

Training and inference scenario

Model training and inference are supported. You can subscribe to model assets (optional), training resources, and inference resources. Subscribed paid or free models can be used for training, and the trained models can then be deployed for inference to achieve better performance.

4

Agent scenario

Subscribe to model assets (optional), training resources (optional), and inference resources. You can use the paid or free models that have been subscribed to in the agent.

Scenario 1: Data scenario (Use only the data engineering capability.)

Subscribe to data general computing unit and data intelligent computing units for data processing, synthesis, labeling, and ratio-based storage. Subscribe to data hosting units for data storage (including datasets and prompt templates).

Figure 1 Data resources

Scenario 2: Inference scenario (Deploy and perform inference using a subscribed paid model or a free model in Model Square.)

Subscribe to ModelArts Studio - Inference Unit (NLP) for the deployment and inference of free and paid NLP models.

Figure 2 Inference scenario

Scenario 3: Training and inference scenario (Subscribe to model assets, training resources, and inference resources to train and deploy models.)

Subscribe to ModelArts Studio - Training Unit, use the subscribed paid model or a free model in Model Square to train the model, and deploy the model after the training is complete.

Figure 3 Training resources

Scenario 4: Agent scenario (Use a deployed model for an agent.)

Subscribe to model assets (optional), training resources (optional), and inference resources. You can use the paid or free models that have been subscribed to in the agent.

Figure 4 Agent

Procedure

  1. Log in to ModelArts Studio Large Model Development Platform using a master account and click Order Now.
  2. In the Development Scenario area, select one or more large models.
    Figure 5 Selecting a development scenario

  3. Subscribe to data resources, training resources, and inference resources by referring to Table 2.
    Table 2 Resource subscription

    Resource Name

    Subscription Item

    Application Scenario

    Billing Mode

    Data resources

    ModelArts Studio - Data Hosting Unit

    Stores data (including datasets and prompt templates).

    Yearly/Monthly

    ModelArts Studio - Data General Computing Unit

    Applies to regular expression operator processing during data processing.

    For details about the data resource types required by different data processing operators, see Introduction to Preset Processing Operators.

    Pay-per-use (duration) or yearly/monthly

    ModelArts Studio - Data Intelligent Computing Unit

    Applies to AI operator processing during data processing.

    For details about the data resource types required by different data processing operators, see Introduction to Preset Processing Operators.

    Pay-per-use (duration) or yearly/monthly

    Training resources

    ModelArts Studio - Training Unit

    Applies to model training and compression of all models.

    Pay-per-use (duration) or yearly/monthly

    Inference resources

    ModelArts Studio - Inference Unit (NLP)

    Applies to the inference service scenario of NLP models on the basic platform and agent development.

    Yearly/Monthly

Renewing Model Assets

ModelArts Studio supports the renewal of model assets in yearly/monthly mode, allowing you to extend the usage period of your subscribed model assets.

To renew a model asset, perform the following steps:

  1. Log in to ModelArts Studio and click Order Management in the upper right corner of the page.
  2. On the Order Management page, click the Model ordering tab. In the subscription model list, click Renew in the Operation column.
  3. On the Renewal Management page, renew the model asset as prompted.

Unsubscribing from a Model Asset

To unsubscribe from a model asset, perform the following steps:

  1. Log in to ModelArts Studio and click Order Management in the upper right corner of the page.
  2. On the Order Management page, click the Model ordering tab. In the subscription model list, click Unsubscribe in the Operation column.

Unsubscribing from a model asset does not affect running tasks such as model training, compression, evaluation, and deployment. However, after the model asset is unsubscribed, the model cannot be selected to create a task. Exercise caution when performing this operation.

Unsubscription is a high-risk operation. Before unsubscribing from a model asset, ensure that you have saved all necessary data and progress to avoid unnecessary losses.

Purchasing Additional Model Assets

In ModelArts Studio, you can purchase additional model assets based on your current subscription. Once the purchase is completed, you can utilize multiple model assets.

To purchase additional model assets, perform the following steps:

  1. Log in to ModelArts Studio and click Order Management in the upper right corner of the page.
  2. On the Order Management page, click New purchase n the upper right corner.
  3. On the Purchase page, purchase new model assets by referring to Procedure.

Renewing Resources

ModelArts Studio supports yearly/monthly subscriptions for data, training, and inference resources.

To renew resources, perform the following steps:

  1. Log in to ModelArts Studio and click Order Management in the upper right corner of the page.
  2. On the Order Management page, click the Resource ordering tab. In the resource list, click Renew in the Operation column.
  3. On the Renewal Management page, renew the resource as prompted.

The pay-per-use billing mode does not support renewal.

Unsubscribing from Resources

To unsubscribe from a resource, perform the following steps:

  1. Log in to ModelArts Studio and click Order Management in the upper right corner of the page.
  2. On the Order Management page, click the Resource ordering tab. In the resource list, click Unsubscribe in the Operation column.

Unsubscription will stop running services, and the resource will be unavailable.

Unsubscription is a high-risk operation. Before unsubscribing from resources, ensure that you have saved all necessary data and progress to avoid unnecessary losses.

Scaling Resources

ModelArts Studio supports scaling of data, training, and inference resources.

To scale resources, perform the following steps:

  1. Log in to ModelArts Studio and click Order Management in the upper right corner of the page.
  2. On the Order Management page, click the Resource ordering tab. In the resource list, click Scaleup in the Operation column.
  3. On the Scaleup page, complete the scaling of the current resource. The platform will pay or refund the price difference based on the specifications before and after the scaling.

Scale-in may affect the creation of ongoing tasks and subsequent tasks. Before scale-in, ensure that the resources to be scaled in have been released.