Performant, secure, AI-friendly unified gateway
API Gateway is divided into Cloud Native API Gateway and AI Gateway. AI Gateway is newly designed for AI scenarios, providing a unified proxy for LLMs and MCP Server.
High cost performance
The service helps you save hidden costs and ensure overall Return on Investment (ROI) based on its performance, ease of use, stability, and security enhancements.
Performance improvement
Compared with self-built gateways, the service has 90% higher overall performance thanks to a software-hardware integrated architecture that combines servers, OSs, and gateways into one to reduce performance loss.
Enhancements
The service also additionally provide features such as out-of-the-box Web Application Firewall (WAF) protection, authentication, protocol conversion, route-level throttling and degradation, service discovery (of various sources), and plug-in marketplace.
High availability
Designs such as multi-AZ deployment and node self-healing guarantee a gateway service level agreement (SLA) higher than 99.95%.
Openness
The service is compatible with mainstream open-source gateways and provides independent open-source Higress solutions. You can switch between the commercial and community editions.
Ease of use
The service provides a complete and continuously evolving observability system that deeply integrates with cloud services and is O&M-free.
Features
Cloud-native API Gateway
Traffic gateway
Adopts the Envoy kernel that is compatible with Kubernetes Ingress standards for ingress traffic management, including route-based forwarding, security authentication, and protocol conversion.
Microservices gateway
Improves end-to-end stability based on support for various service sources (such as K8s and Nacos) and microservices governance features (such as end-to-end canary release and throttling-based degradation).
API management
Provides API hosting services throughout lifecycle, covering stages of API design, development, testing, publishing, sales, O&M, security control, and unpublishing.
Original API Gateway
Large Model API Proxy
Unifies the proxy for mainstream large models and self-built large model services, providing OpenAI-compatible access methods. It offers secondary API key issuance, rate limiting, security protection, observability, and other governance capabilities.
MCP Management
Supports rapid conversion from API to MCP and provides MCP Server proxy, security authentication, as well as unified observability and rate-limiting governance capabilities.
Scenarios
AI Application Traffic Entry and Integration
Aimed at LLM, it unifies the proxy for mainstream large models and self-built model services, providing OpenAI-compatible access methods, as well as secondary API KEY issuance, rate limiting, security protection, observability, and governance capabilities.
What it can do
-
High Availability
Multi-model gray release, failover between multiple models, token quota management, and flow control.
-
High Security
Provides content safety checks for large models, supports routing configuration authentication and consumer authorization, enabling control, security, and policy management of API access.
-
Reduce Model Hallucination
Provides online search, automatic fusion of search results, and intent recognition for questions.
-
Enterprise-level Observability
Enables OT services for full-link tracking in applications, gateways, and backend LLM services, connecting logs, request parameters, and other information through TraceId.
Unified Management and Access for MCP Server
Aimed at MCP Server, it provides MCP Server proxy, security authentication, as well as unified observability, rate limiting, and other governance capabilities.
What it can do
-
API-to-MCP
Supports the direct conversion of REST API into MCP Server, avoiding the repetitive work of rebuilding and maintaining the MCP Server.
-
Protocol Offloading
Seamlessly supports the latest official MCP protocol, reducing upgrade costs; for example, supports converting SSE to Streamable HTTP, avoiding the need for stateless applications to use SSE.
-
MCP Marketplace
Provides an officially maintained MCP marketplace, ensuring that the MCP server is usable, user-friendly, and secure.
Service exposure and traffic control
Cloud-native API Gateway is easy to integrate, use, and scale and hot-updatable. It supports K8s Ingress standards, backend services such as microservices and functions, and discovery methods such as K8s and Nacos.
What it can do
-
Traffic management for containers and microservices
As the access layer of Container Service for Kubernetes (ACK), Cloud-native API Gateway is compatible with the Kubernetes Ingress standard and provides fine-grained routing and API control. The service supports one-click import of K8s services or Nacos-registered services and direct use of pod IP addresses for communications. In addition, extensions of core annotations of NGINX Ingress are supported and capabilities such as throttling, warm-up, and canary release are provided.
-
Serverless
The service works with Function Compute to provide an ideal serverless computing platform. One function corresponds to one API to quickly provide services based on powerful and easy-to-use authentication and flow control capabilities.
-
Multi-active disaster recovery
It functions as a unified access layer for multiple clusters to implement multi-active disaster recovery and other features. The service supports multiple service sources, such as Alibaba Cloud ACK, MSE Nacos, Function Compute, Domain Name System (DNS). It can also accurately route requests by proportion or request content and supports health check and service fallback.
Full-lifecycle API management
This service manages APIs throughout their lifecycles from design and development to publishing and unpublishing. In addition, it supports multiple environments to help teams manage and publish APIs more efficiently and securely, and in turn promotes rapid business iteration and innovation.
What it can do
-
Standardized development for efficiency boosting
This service manages APIs throughout their lifecycles from design and development to publishing and unpublishing. In addition, it supports multiple environments to help teams manage and publish APIs more efficiently and securely, and in turn promotes rapid business iteration and innovation.
-
Fine-grained API management
In scenarios where upstream systems open capabilities to downstream partners or internal systems interconnect with external ones for collaboration, APIs can be managed in a fine-grained manner to control the scope of opening.
-
Multi-environment management
This service supports multiple mutually isolated environments. This way, operations in the development or test environment do not affect business in the production environment and tests or verifications can be performed at different development stages.
Specifications
AI Gateway
Unified Proxy for Large Model API and MCP Server
Features
-
![]()
Unified proxy for large model API, high availability, high security, and reduced model hallucination.
-
![]()
Unified management and access for MCP Server, API-to-MCP, and protocol offloading.
-
![]()
Provides an officially maintained MCP marketplace.
Unified proxy for large model API, high availability, high security, and reduced model hallucination.
Billing method
Supports purchase by resource plan and by the pay-as-you-go billing method. Fees are charged based on the engine specifications and the number of nodes.
Cloud-native API Gateway
Integrates API management, traffic gateway, and microservices gateway capabilities.
Features
-
![]()
Provides API lifecycle services.
-
![]()
Uses Envoy core, compatible with K8s Ingress standards.
-
![]()
Connects multiple service sources such as K8s/Nacos.
Billing method
Supports purchase by resource plan and by the pay-as-you-go billing method. Fees are charged based on the engine specifications and the number of nodes.
