Configure enterprise inference services

Background

On the public cloud, medium and large enterprises increasingly require personalization. They use Supervised Fine-Tuning (SFT) to tailor models for their specific business scenarios. To meet this need, Alibaba Cloud Model Studio provides a one-stop solution for data preparation, model fine-tuning, and model deployment. Qoder CN integrates with Alibaba Cloud Model Studio to help enterprises efficiently meet their needs for model fine-tuning and personalization.

Prerequisites

1. Service purchase

   • You have activated the model training service on Alibaba Cloud Model Studio. If not, go to the Alibaba Cloud Model Studio console to activate it.

   • You have purchased Qoder CN Enterprise Edition (Enterprise Standard Edition or Enterprise Dedicated Edition). For more information, see Quick Start for Enterprise Standard Edition and Quick Start for Enterprise Dedicated Edition.

2. Get the inference service API

   1. Deploy the SFT model on Alibaba Cloud Model Studio or other supported model service platforms. For more information, see Introduction to model deployment.

   2. Generate an API URL for the inference service and record key information, such as the API URL, model name, and API key. For more information, see Make your first call to the Qwen API.

After an enterprise administrator completes the API configuration for their own inference service, the Qoder CN system must ensure that when the enterprise authorizes developers to use Qoder CN services, it automatically calls the model service configured within the enterprise.

Procedure

1. Log on to the Qoder CN console. In the left-side navigation pane, choose Policy Configuration > Inference Service Configuration.

2. On the Inference Service Configuration page, configure the model service for your required scenario:

   • Turn on the switch for the model service configuration.

   • Enter the model URL, model name, and API key.

3. Click Test Now.

   1. Success: The model service is connected.

   2. Failure: The model service connection failed. Verify your configuration and test again.

4. Click Save.

5. In your IDE, log in to your enterprise account. In the intelligent session window, select a model service from the input field to see your configured models. Available options include models such as qwen3, qwen3-thinking, qwen2.5-max, deepseek-r1, and testmodel(org).

Disclaimer