Platform For AI:Configure connections

Connection type

Key parameter description

General LLM Service

Configure a connection to an LLM service deployed on EAS or a third-party platform. For more information about deployment, see Model deployment and training.

• Service Provider:

  • PAI-EAS model service: Use a PAI-EAS model service. For EAS Service, select your deployed LLM service. The base_url and api_key are automatically set to the service's virtual private cloud (VPC) endpoint and token, respectively.

  • Third-party model service: Use a third-party model service. For example, for the official DeepSeek service, set the base_url to https://api.deepseek.com and obtain the api_key from the DeepSeek official website.

• Model Name: When you deploy a model from Model Gallery as an EAS service, you can find the model name on the model details page. To access this page, click the model card on the Model Gallery page.

  

• Tool Call: Indicates whether the model supports the Tool Call feature of the OpenAI API. When using an Agent node, you must select a model that supports Tool Call.

General Embedding Model Service

Configure a connection to an embedding model service deployed on EAS or a third-party platform. For more information, see Model deployment and training. The configuration method is the same as for the General LLM Service connection.

General Multimodal Embedding Model Service

Configure a connection to a multimodal embedding model service deployed on EAS or a third-party platform. For more information, see Model deployment and training. The configuration method is the same as for the General LLM Service connection.

Model Studio Service

Configure a model service connection for the Alibaba Cloud Model Studio platform. Key parameter:

api_key: Go to Alibaba Cloud Model Studio - API KEY to create or find the API key.

DeepSeek Model Service

Configure a connection for the DeepSeek model service. Key parameter:

api_key: Go to the DeepSeek official website to create or find your API key.

AI Search Open Platform Model Service

Multimodal VL Intelligent Annotation Model

Use this to intelligently annotate multimodal data. Only model services deployed on EAS are supported.

Connection type

Key parameter description

Search and Analytics Service - Elasticsearch

• url: The endpoint of the Elasticsearch instance, in the format http://<private_endpoint>:<private_port>. You can find the private endpoint and private port in the Basic Information section of the target Elasticsearch instance's details page in the Elasticsearch console.

• username/password: The default username for Elasticsearch is elastic. The password is the logon password you set when you create an Elasticsearch instance. If you forget the logon password, you can reset the instance access password.

Vector Database - Milvus

• uri: The endpoint of the Milvus instance, in the format http://<Milvus_internal_endpoint>. For example, http://c-b1c5222fba****-internal.milvus.aliyuncs.com. You can find the Milvus internal endpoint in the Endpoint section of the target instance's details page in the Milvus console.

• token: The username and password for logging on to the Milvus instance, in the format <yourUsername>:<yourPassword>.

• database: The database name. When you create a Milvus instance, the system creates a default database named `default`. You can also manually create a new database.

Hologres - Real-time Data Warehouse

• host/port: Go to the Hologres console. On the details page of the target instance, find the host and port for the specified VPC in the Network Information section.

• database: The database name set when you create a database.

• user/password: Go to the details page of the target Hologres instance. On the Account Management page, create a custom user.

  For Select Member Role, select Instance Super Administrator (SuperUser).

DashVector Vector Database

• endpoint: Go to the DashVector console. On the details page of the target cluster, find the virtual private cloud (VPC) endpoint in the Access Port section.

• api_key: To get the API key, see Manage API keys.

ApsaraDB RDS for MySQL

• host/port: To get the host (internal endpoint) and port (internal port), see View and manage instance connection addresses and ports.

• database: The database name set when you create a database.

• user/password: The account and password set when you create a database account.

Connection type

Key parameter description

Function AI

Configure an MCP service deployed through Alibaba Cloud Function AI.

Select an MCP service from the drop-down list to automatically fill in the connection parameters. The url defaults to the public endpoint of the MCP service.

Custom

Configure an MCP service that you deployed or that is hosted on a third-party platform.

• transport: The underlying transport protocol used to communicate with the MCP service. Supported protocols are SSE and StreamableHTTP.

• url: The full endpoint URL of the MCP service.

• authentication: The identity authentication information for accessing the MCP service. The following three methods are supported:

  • BearerAuth: Uses Bearer token authentication. You must enter a valid access token in the token field.

  • Other: You must specify the auth_header_name and token to indicate which request header is used to obtain the token.

  • None: No identity authentication is required.