All Products
Search

This Product

    DataWorks:Deploy models

    Document Center

    DataWorks:Deploy models

    Last Updated:Oct 23, 2025

    Deploy pre-trained LLMs on serverless infrastructure. No cluster management required.

    Billing

    Model deployments incur charges from DataWorks and related Alibaba Cloud services:

    Service

    Charges

    DataWorks fees

    Billed per CU-hour when models are running. For more information, see Billing of serverless resource groups.

    Non-DataWorks fees

    How it works

    DataWorks runs your model as a fully managed service inside a serverless resource group. For secure internal access, the system automatically configures PrivateLink and Private Hosted Zone during deployment:

    • PrivateLink: Creates an encrypted tunnel between your VPC and the DataWorks resource group, enabling private cross-VPC communication.

    • Private Hosted Zone: Provides DNS resolution so you can reach your model using a custom domain name from within your VPC.

    API calls using the model's domain name travel through the PrivateLink tunnel from your VPC to the model instance running in the DataWorks resource group. You can monitor these resources in the PrivateLink console and the Alibaba Cloud DNS console.

    Usage notes

    Important

    All services must be in the same region as your workspace.

    Limitations

    • Regions: China (Hangzhou), China (Shanghai), China (Beijing), China (Ulanqab), China (Shenzhen), China (Hong Kong), Japan (Tokyo), Singapore, Malaysia (Kuala Lumpur), Indonesia (Jakarta), Germany (Frankfurt), US (Silicon Valley), and US (Virginia).

    • Deployment: Models only (training not supported).

    • Quotas: 50 services per Alibaba Cloud account per region, 5 services per resource group.

    • Networking: 3 VPCs maximum per service.

    Deploy a model

    Access the deployment page

    1. Log on to the DataWorks console and switch to your target region.

    2. Click Model Service in the left navigation pane.

    3. Click Deployment Model, select a model, then click Deployment.

    Configure the deployment

    • Model: The model to deploy. For more information, see Manage model services.

    • Service Name: A unique identifier for this deployment.

    Resource settings

    Setting

    Description

    Resource Group

    Your serverless resource group. View the usage in the console under Resource Group > Resource Group Name > Details.

    Switch

    A vSwitch attached to your resource group.

    Deployment Specifications

    Select a resource specification.

    Number of instances

    Multiple instances improve availability.

    Total occupancy

    Calculated automatically: Deployment Specifications × Number of instances.

    Ensure your resource group has enough available CUs for the model service. For more information, see Allocate CU quotas.

    • Subscription groups: Expand capacity as needed.

    • Pay-as-you-go groups: 500 CU default, auto-expanded to 2000 CU after first deployment.

    Click Deploy.

    Important

    DataWorks automatically creates PrivateLink, Private Hosted Zone, and security groups. Don't delete or modify these. They're removed automatically when you delete the service.

    After deployment completes:

    • A PrivateLink endpoint appears in your default VPC, connected to the DataWorks resource group.

    • A DNS record is added to Private Hosted Zone, mapping the service domain to the PrivateLink endpoint in your VPC.

    Manage models

    After a model service is created, you can manage the model status, view service information, manage network and API keys, and adjust resources as needed from the Model Service list.

    Manage the model status

    A model service starts by default after creation. You can manage its status in the Actions column.

    Operation

    Model service status change

    Resource consumption

    Start

    In operation

    Consumes serverless resource group resources.

    Stop

    Stopped

    Does not consume serverless resource group resources.

    Delete

    -

    Completely releases serverless resource group resources.

    View the model service

    The Overview tab displays the configuration information for the current model service.

    1. In the Model Service list, find the target model service and click its name to go to the Overview tab.

    2. You can manage the model service's Basic Information, Resource Allocation, and Invocation Information.

      • Basic Information: Such as the model service name, service ID, and type.

      • Resource Allocation: Includes details such as Deployment Specifications and Number of instances.

      • Invocation Information: To invoke the service and use the model in a node task, click the image icon next to VPC Address Invocation Domain Name to copy and obtain the domain name parameter.

    Modify model resources

    For a created model service, you can modify the service name, adjust the deployment specifications, and change the number of instances for deploying the model service.

    1. In the Model Service list, click the target model service name to go to the Overview tab.

    2. Click Modification next to Resource Allocation and configure the settings.

      Important

      Changing resources causes the service to restart, which affects the operation of the model service.

    Manage the model network

    The Network Configuration tab displays the VPCs that can currently access the model service through the internal network. On the tab, you can add or manage VPCs that can be used to access the model service.

    1. In the Model Service list, click the target model service name to go to the Overview tab.

    2. Switch to the Network Configuration tab to view the VPCs.

    3. To expand the access range, you can click Add Network to allow more VPCs to access the model service deployed on DataWorks.

      When adding a network, you must specify a VPC and a vSwitch. You can access the model service through the VPC after its status changes to Available.
      Note

      • Billing: After adding a VPC for the model service, the system creates a PrivateLink endpoint in the VPC you selected to access the model service, establishing network connectivity with the DataWorks resource group. A DNS record is also added to Private Hosted Zone. This process incurs instance fees, traffic processing fees, and domain name resolution fees. For more information, see Billing of PrivateLink and Billing of Private Hosted Zone.

      • Limit: You can add a maximum of three VPCs for each model service.

    4. If you no longer want to allow a certain VPC to access the model service, you can click Delete for the target VPC in the model service.

      When you delete the VPC from the model service, the PrivateLink endpoint created in that VPC is also removed.

    Manage API keys

    An API key is an authentication credential provided by the model service to authenticate the caller's identity and permissions. You can manage all API keys for invoking the current model service on the API Key tab.

    1. In the Model Service list, click the target model service name to go to the Overview tab.

    2. Switch to the API Key tab to create, manage, and use API keys.

      • Add API Keys: After the model service is deployed successfully, DataWorks auto-generates an internal key for platform integrations. If you need to call the service model through the model service Endpoint in other environments, click Add New API Key to create a new API key.

        It is recommended that you create separate API keys for different use cases.

      • View API Keys: Click View in the Actions column of the target API Key, then click Copy to obtain the API Key.

      • Delete API keys: DataWorks API Keys provide Disable and Delete functions.

        Important

        If you need to disable or delete an enabled API key, evaluate the impact in advance. Once an API key is disabled or deleted, all tasks that use the API key to invoke models will fail.

        Disable or Delete operation takes approximately 5 min to take effect.

    More operations

    After deploying the model, you can use the model to develop related tasks.

    How invocation works

    The following descriptions explain how model service invocation works:

    image

    When you deploy a model service in a DataWorks resource group or configure a VPC for it, the system automatically performs the following operations:

    1. Establish a cross-VPC connection. In your VPC (the VPC in your account that can communicate with the DataWorks resource group), the system automatically creates a PrivateLink endpoint and establishes an encrypted communication channel with the PrivateLink endpoint service in the DataWorks resource group VPC.

      This operation automatically creates a PrivateLink endpoint in your account. Your account must have the PrivateLink service enabled.

    2. Configure domain name resolution service. The system automatically configures domain name resolution rules in the VPC, so that domain name request traffic within the VPC is automatically forwarded to the DataWorks model service.

      This operation automatically deploys the Private Hosted Zone service in your account. Your account must have this service enabled.