Learn about the service deployment scope, endpoints, and supported features of the China (Beijing) region.
Region overview
The region determines the access point and data storage location. Choose a nearby region to reduce latency.
- Service deployment scope: determines the inference execution location.
- Endpoint: affects concurrency limits, timeouts, and other service guarantees. Each region has independent endpoints.
- The application sends a request to the selected region through the endpoint, and the request data is stored in that region.
- The access region forwards the request to an inference node within the service deployment scope for computation. Transmission is encrypted end-to-end.
- The inference result returns to the access region for storage and is then returned to the application. User static data is always stored in the selected region.
| Region | Service deployment scope |
|---|---|
| China (Beijing) | China Elastic Zone: Inference can run in the Chinese mainland and other regions where Model Studio is available, excluding the United States, European Union member states, Japan, and South Korea. Data security is maintained throughout resource scheduling. |
| Chinese mainland: Inference runs only in the Chinese mainland. This scope applies to models launched in China (Beijing) before October 13, 2026; models launched on or after this date use China Elastic Zone by default. |
Endpoints
Model Studio provides three types of endpoints for model inference APIs in the China (Beijing) region: workspace-specific endpoints, DashScope endpoints, and trial endpoints. These are suitable for scenarios ranging from trial experiences to enterprise-level production. Workspace-specific endpoints are recommended. The key differences are as follows:
Item | Workspace-specific endpoint (recommended) | DashScope endpoint | Trial endpoint |
|---|---|---|---|
Endpoint format |
|
|
|
Applicable scenario | Recommended for production environments. Provides higher concurrency capacity and network isolation, ensuring stable, low-latency access in high-traffic scenarios. | For existing business compatibility. We recommend migrating to the workspace-specific endpoint. | For quick trials and functional validation. Not recommended for production environments. |
Authentication method | Access only the current workspace | Access all workspaces | Access all workspaces |
Rate limit | RPM and TPM vary by model and are aggregated at the Alibaba Cloud account level | RPM and TPM vary by model and are aggregated at the Alibaba Cloud account level | RPM is 1000 (at the Alibaba Cloud account level); TPM varies by model |
Request timeout | 3600 seconds | 600 seconds | 600 seconds |
Protocol support | HTTP, SSE, WebSocket, WebRTC, AOQ (AI over QUIC) | HTTP, SSE, WebSocket | HTTP, SSE |
SLA | 99.9% | 99.9% | Not provided |
Access information
The China (Beijing) region has independent endpoints, API keys, and model lists. These cannot be mixed across regions.
Region | Region ID | Workspace-specific endpoint | DashScope endpoint | Trial endpoint | API Key | Model list |
|---|---|---|---|---|---|---|
China (Beijing) |
|
|
|
|
Migrate to workspace-specific endpoints
Migrating from a DashScope endpoint or trial endpoint to a workspace-specific endpoint requires only two steps, with no need to modify business logic code:
- Obtain the workspace-specific endpoint: On the Workspace Management page, copy the content in the API Host column.
-
Replace the domain name in the request address: Replace the original domain name in your code with the copied API Host (for example,
llm-xxx.cn-beijing.maas.aliyuncs.com).llm-xxxis the workspace ID. The request addresses for APIs in the China (Beijing) region are as follows:- OpenAI-compatible API: replace
https://dashscope.aliyuncs.com/compatible-mode/v1withhttps://llm-xxx.cn-beijing.maas.aliyuncs.com/compatible-mode/v1 - DashScope API: replace
https://dashscope.aliyuncs.com/api/v1withhttps://llm-xxx.cn-beijing.maas.aliyuncs.com/api/v1 - Anthropic-compatible API: replace
https://dashscope.aliyuncs.com/apps/anthropicwithhttps://llm-xxx.cn-beijing.maas.aliyuncs.com/apps/anthropic
- OpenAI-compatible API: replace
Feature support
Region | Real-time inference | Batch inference | Model experience | Model monitoring (basic) | Model monitoring (advanced) | Model alerting | Transmission security | Permission management | Model fine-tuning |
|---|---|---|---|---|---|---|---|---|---|
China (Beijing) | Supported | Supported | Supported | Supported | Supported | Supported | Supported | Supported | Supported |
Related documents
- Tongyi Qianwen API reference
- Recommended models — Models and context lengths by region
- Model inference pricing — Pricing by region /China (Beijing)/
- Rate limiting— RPM and TPM limits
- Obtain an API key — Create and manage API keys
- Base URL overview — Model service endpoints
- Regions and endpoints - More region information