Eden AI
Unified API aggregator for AI services across providers
Higress
Cloud-native API gateway with AI and Wasm extensibility
Higress is a cloud-native API gateway built on Istio and Envoy, offering AI model routing, MCP server hosting, and extensible Wasm plugins for traffic, security, and observability.
Overview
Overview
Higress delivers a production‑grade API gateway that combines the robustness of Istio/Envoy with AI‑focused capabilities. It unifies access to major LLM providers, supports the Model Context Protocol (MCP) for tool integration, and provides a rich UI console for management and observability.
Deployment & Extensibility
Developers can launch a full instance with a single Docker command or deploy via Helm in Kubernetes. The gateway’s plugin architecture relies on WebAssembly, allowing safe, hot‑updatable extensions written in Go, Rust, or JavaScript. Built‑in plugins cover traffic management, security (WAF, auth), and AI‑specific functions, meeting most enterprise scenarios out of the box.
Target Audience
Ideal for teams building AI agents, microservice platforms, or secure ingress points that need high‑throughput, low‑latency streaming and fine‑grained rate limiting. Enterprises benefit from Alibaba’s production validation and optional enterprise support.
Highlights
Production‑grade performance validated by Alibaba at massive scale
True streaming processing for request/response bodies
Wasm‑based extensibility supporting Go, Rust, and JavaScript
Built‑in UI console, auto TLS via Let’s Encrypt, and Docker‑first start
Pros
- High throughput and low latency for AI workloads
- Memory‑safe plugin isolation with hot updates
- Unified management of LLM and MCP APIs
- Comprehensive security and observability out of the box
Considerations
- Steep learning curve for Istio/Envoy newcomers
- Advanced features may require the enterprise edition
- Kubernetes deployment adds operational complexity
- Documentation may be fragmented across multiple repos
Managed products teams compare with
When teams consider Higress, these hosted platforms usually appear on the same shortlist.
Looking for a hosted option? These are the services engineering teams benchmark against before choosing open source.
Fit guide
Great for
- Enterprises needing a reliable AI gateway with load balancing
- Developers building AI agents that call external tools via MCP
- Teams requiring a customizable microservice ingress controller
- Organizations prioritizing security, rate limiting, and observability
Not ideal when
- Simple static websites without API requirements
- Projects that cannot adopt Kubernetes or Istio
- Use cases demanding a lightweight proxy without Wasm
- Environments that require proprietary vendor lock‑in
How teams use it
MCP Server Hosting
Host remote MCP servers via Wasm plugins, enabling AI agents to invoke tools with unified auth, rate limiting, and audit logging.
AI Model Routing
Connect to multiple LLM providers through a single gateway, applying token rate limits, caching, and multi‑model load balancing.
Kubernetes Ingress Controller
Replace Nginx ingress with Higress, leveraging familiar annotations while gaining AI‑aware traffic management.
Security Gateway
Enforce WAF, JWT, OIDC, and other authentication strategies with built‑in plugins and automatic TLS management.
Tech snapshot
Go76%
C++18%
Rust3%
Shell1%
Makefile1%
TypeScript1%
Frequently asked questions
How do I start Higress locally?
Run the provided Docker command; it pulls the all‑in‑one image, writes config to the current directory, and exposes UI (8001) and gateway ports (8080/8443).
Which LLM providers are supported?
Higress supports all mainstream model providers, both domestic and international, via a unified protocol.
Can I write my own plugins?
Yes, you can develop Wasm plugins in Go, Rust, or JavaScript and load them through the plugin mechanism.
Is enterprise support available?
Enterprise‑level support is offered by contacting the project maintainer via the listed email or social media accounts.
Do I need Kubernetes to use Higress?
Kubernetes is optional; Higress can run as a standalone Docker container, though Helm charts are provided for K8s deployments.
Project at a glance
Active- Stars
- 7,682
- Watchers
- 7,682
- Forks
- 1,012
LicenseApache-2.0
Repo age3 years old
Last commit3 days ago
Primary languageGo
Last synced 2 hours ago