Help Center/ PanguLargeModels/ User Guide/ Developing a DeepSeek Model/ Deploying a DeepSeek Model
Updated on 2025-11-04 GMT+08:00
View PDF
Share

Deploying a DeepSeek Model

After the model training is complete, you can deploy the model.

  1. Log in to ModelArts Studio and access the required workspace.
  2. In the navigation pane, choose Model Development > Model Deployment. Click Create Deployment Task in the upper right corner.
  3. In the Select Model dialog box, set Sources to Model Square and Type to NLP, select a model, and click OK.
  4. On the Create a deployment page, set deployment parameters by referring to Table 1 and start model deployment.
    Table 1 Parameters for deploying a third-party model

    Category

    Parameter

    Description

    Deployment Configuration

    Select Model

    You can modify the following information:

    • Sources: Select Model Square.
    • Type: Select NLP and select the model and version to be deployed.

    Deployment Mode

    Deployment at the online: Algorithms are deployed in the resource pool provided by the platform.

    Deployed_model

    Unique identifier of the inference service when the inference service is called through the inference API of the V2 version.

    Safety Fence

    Open and agree to authorize

    Ensures the security of model calling.

    Version Selection

    Currently, only the basic edition of the security guardrail is supported, which is built-in with default content moderation rules.

    Resource Configuration

    Billing model

    Free for a limited time

    Instance Count

    Set the number of instances required by the model to be deployed. It is recommended that the number of instances to be deployed at a time be less than or equal to 10. Otherwise, traffic limiting may be triggered and the deployment may fail.

    Safety Fence

    Enabled/Disabled

    In the trial version, the security fence switch is enabled by default to effectively block harmful content during model inference.

    Basic information

    service name

    Set the name of the deployment task.

    Description (Optional)

    Set the description of the deployment task.

    Tag (Optional)

    Sets the tag of a deployment task. When deploying or updating a model, use SERVICE_TAGS_CONFIG as the key and the Base64 character string of the JSON body of the tag information as the value to add the tag to the environment variables of the model service.
  5. After setting the deployment parameters, click Deploy Now.