Model Context Protocol (MCP)
MCP eliminates the need to write complex interfaces for each external tool, enabling Alibaba Cloud Model Studio applications to connect to a vast ecosystem of third-party tools.
MCP service integration examples
Route planning agent | Web scraping workflow |
|---|---|
|
|
|
|
How to use MCP
Alibaba Cloud Model Studio agent and workflow applications now support two types of MCP services. For detailed instructions, see Official MCP services and Custom MCP services.
- Official MCP services: Alibaba Cloud Model Studio offers several official MCP services that you can readily connect to your applications.
- Custom MCP services: You can find MCP services in the MCP marketplace or on the internet and deploy them as custom services on Alibaba Cloud Model Studio.
Billing
Cloud-deployed MCP services
- Deployment fee: Free for a limited time.
- Invocation fee: Some MCP services may incur fees for third-party API calls. These fees are charged by the third-party provider, not by Alibaba Cloud Model Studio.
Web search MCP service
- Service activation: To activate and use this service, go to Web Search MCP Service.
- Invocation fee: This service includes a free tier of 2,000 invocations. After the quota is used up, invocations are billed at CNY 29 per 1,000 invocations.
- Throttling: The throttling limit is 15 QPS, shared between an Alibaba Cloud account and its RAM users.
Custom-deployed MCP services
Alibaba Cloud Model Studio offers the following two billing modes based on service response time:
-
Basic mode:
-
No deployment fee. Billing is based on invocation duration, with no charges when the service is idle.
- Invocation rate: CNY 0.000156 per second
-
Use cases: Suitable for applications with infrequent calls that are not sensitive to startup latency. The first invocation might experience a cold start delay.
-
No deployment fee. Billing is based on invocation duration, with no charges when the service is idle.
-
High-speed mode:
-
You are billed for the deployment duration. When the service is invoked, you are also billed for the invocation duration.
- Invocation rate: CNY 0.000156 per second
- Deployment rate: CNY 0.000036 per second
-
Use cases: Ideal for applications that require high availability and frequent invocations, as it reduces cold start latency.
-
You are billed for the deployment duration. When the service is invoked, you are also billed for the invocation duration.

