Higress is an AI-native API gateway that began inside Alibaba and is now an open-source project on GitHub with more than 9,600 stars and frequent chart appearances. Built on Envoy and Istio, it has evolved from a cloud-native ingress controller into a gateway designed specifically for LLM traffic, with updates continuing through 2026.
As applications call more external models, they need one control point for routing, security and cost management. Higress provides that layer: it proxies requests to many LLM providers behind a single OpenAI-style endpoint, applies token-based rate limiting and quotas per consumer, caches responses, and adds safety checks on prompts and outputs. A rich Wasm plugin system lets teams extend behavior without forking the gateway, and because it descends from production ingress technology, it handles classic API gateway duties, Kubernetes ingress and microservice routing in the same deployment.
Higress itself is free and open source under the Apache License 2.0. Alibaba Cloud also offers commercial cloud gateway services and enterprise support built on Higress for teams that want a managed option, making the overall offering Freemium.
Higress is for platform teams and AI application developers who need to govern traffic to multiple LLMs with quotas, caching and safety controls, especially in Kubernetes environments. It is a practical choice for organizations standardizing on one gateway for both conventional APIs and AI traffic.