AI Gateway Serverless (new) is a Serverless edition of API Gateway that is entering public preview. It provides a lower-barrier entry point for AI application traffic management, covering scenarios such as model routing, Agent API, MCP Server, AI security protection, and token usage analytics.
Public preview period
The public preview for AI Gateway Serverless (new) runs from July 31, 2026 to August 31, 2026.
During the public preview period, you can create and use AI Gateway Serverless (new) instances in supported regions. For the exact availability timeline, purchasable instance types, and feature scope, refer to the API Gateway console.
Supported regions
The following regions are supported in this public preview:
China (Beijing)
China (Shanghai)
China (Hangzhou)
China (Shenzhen)
China (Hong Kong)
Regions will expand after testing. Check the console for updates.
Preview editions
This public preview covers the following AI Gateway Serverless (new) instance types:
Instance type | Use case |
Serverless Standard | Quick trials, PoC, development and testing, and low-traffic workloads. |
Serverless Enterprise | Production workloads that require dedicated entry capabilities or custom domain names. |
If you are already using Serverless (legacy) instances, those instances remain available. For feature differences and edition selection guidance, refer to the product documentation and the API Gateway console.
Key capabilities
During the public preview period, you can use the following capabilities with AI Gateway Serverless (new):
LLM API management — Centrally manage the creation, publication, deprecation, and deletion of model APIs.
Agent API management — Provide a unified entry point and governance for Agent access scenarios.
MCP Server — Support MCP-related integration and proxy scenarios.
AI security protection — Provide content security protection for AI traffic.
Observability and analytics — View token usage, access logs, and AI access statistics.
Traffic management — Support gateway policies such as timeout, retry, QPS rate limiting, concurrency limiting, CORS, and header modification.
For available features, feature limits, and regional support, refer to the API Gateway console.
Billing during public preview
During the public preview period, AI Gateway Serverless (new) gateway fees are waived. Only inbound internet traffic is billed through CDT (Cloud Data Transfer).
Charge type | Billing rule during public preview |
AI Gateway Serverless (new) gateway fees | Waived |
Inbound internet traffic | Billed through CDT |
Other cloud product fees | Charged according to the billing rules of the corresponding products |
Fees for model services, Simple Log Service (SLS), Cloud Monitor (CMS), CDT, and other cloud products are not covered by the AI Gateway Serverless (new) public preview waiver. Actual charges are subject to your bills and the billing rules of the corresponding products.
How to join the public preview
Before you start, confirm the following:
Whether your target region is within the public preview scope.
Whether your workload requires public network access and may generate inbound internet traffic.
Whether you need production-grade capabilities such as custom domain names, certificates, TLS policies, or HTTP/2.
Whether you also use model services, SLS, CMS, or other cloud products that incur separate fees.
During the public preview period, log on to the API Gateway console, select a supported region, create an AI Gateway Serverless (new) instance, and complete the configuration as guided.
Notes
The feature scope, regional availability, and billing incentives during public preview may change based on the product release schedule. Refer to the product console, official announcements, and pricing page.
Gateway fees are waived during public preview, but inbound internet traffic is still billed through CDT.
Other cloud product fees incurred during public preview are charged according to the billing rules of the corresponding products.
After public preview ends, AI Gateway Serverless (new) may enter the commercial billing phase. For commercial charges, pricing, and promotional policies, refer to official announcements, the product console, and the pricing page.