Skip to content

Deploy a Model

You can deploy models from the Model Store or Model Services. Below are the details for configuring the deployment:

  • Select the Model
    Choose the model to deploy (e.g., Qwen2-0.5B-Instruct), ensuring it aligns with your business needs and task requirements.

  • Model Service Name Requirements
    The name of the model service must meet the following criteria:

    • Length: 2–64 characters
    • Allowed characters: Lowercase letters, numbers, and hyphens (-). The name must start and end with a lowercase letter or number.
    • Example: text-gen-service or model-01

  • Region

Select the region for service deployment (e.g., "Shanghai Area 2"). Choose a region based on your business coverage and latency needs.

  • Number of Instances

    • Configure the number of instances to deploy. Default value: 1.
    • More instances enhance service concurrency, but they will also increase costs.

  • Billing Method

    d.run offers two billing options:

    1. Pay-as-you-go:

      • Real-time billing based on usage, ideal for short-term or variable-demand users.
      • Cost formula: Number of instances × hourly rate.
      • Example: 1 instance × 3 Yuan/hour = 3 Yuan/hour.

    2. Annual or Monthly Subscription (currently not supported):

      • Prepay for service instances at a discounted rate, suitable for long-term users.
      • Upon selecting this option, the system will display the corresponding annual or monthly fee.

  • View Configuration Costs

    At the bottom of the page, the cost calculation formula and estimated costs will be displayed automatically.

    • Configuration cost: 3 Yuan/hour
    • Calculation formula: 1 instance × 3 Yuan/hour = 3 Yuan/hour