Kubernetes Gateway API control plane with a Rust data plane
Standard Gateway API routing · Go control plane · gRPC/xDS · Experimental AI and Wasm extensions
Next-generation API gateway for the cloud-native era, combining Go ecosystem with Rust performance
The Go control plane watches Kubernetes cluster state, translating Gateway, HTTPRoute, and Service resources into an internal routing model. The Rust data plane receives runtime snapshots via gRPC/xDS and handles every request with zero GC pauses. Both planes scale independently.
Full implementation of Kubernetes Gateway API v1.5.1 with 55 declared conformance features. Supports GatewayClass, Gateway, HTTPRoute, GRPCRoute, TLSRoute, and more. API-compatible with Envoy Gateway, Istio, and other major implementations for minimal migration cost.
The Rust data plane delivers tens of thousands of QPS per core with no GC and zero-cost abstractions. Memory usage is 60-80% lower than traditional Go/Java gateways, with millisecond startup times. Built-in connection pooling, zero-copy forwarding, and async I/O are battle-tested in production.
Built-in AIService and TokenPolicy resources unify API endpoints and credentials across multiple model providers (OpenAI, HuggingFace, etc.). Token-based rate limiting and quota management provide a single traffic entry point for AI applications, reducing multi-model integration complexity.
WasmPlugin resources support URL, ConfigMap, or inline module sources. Plugins bind to specific Gateways or Routes for custom authentication, traffic coloring, request rewriting, and more — hot-loadable without restarting the gateway process.
Admin API for runtime configuration queries and debugging. Built-in Prometheus metrics endpoints and Grafana dashboard templates. Health and readiness probes. One-click Helm Chart deployment with flexible Kustomize overlays for environments from bare metal to Kubernetes.
Deploy in your Kubernetes cluster with a single Helm command. Gateway API standard, compatible with existing toolchains, up and running in 5 minutes.
View Quick Start →One Helm command to install, one HTTPRoute resource to route traffic
helm repo add nantian-gw https://chart.nantian.dev
helm repo update
helm install nantian-gw nantian-gw/nantian-gw --namespace nantian-gw --create-namespace
apiVersion: gateway.networking.k8s.io/v1
kind: HTTPRoute
metadata:
name: my-app
spec:
parentRefs:
- name: nantian-gw
rules:
- matches:
- path:
type: PathPrefix
value: /api
backendRefs:
- name: my-service
port: 8080
Split-plane control plane and data plane communicating over gRPC xDS
One platform for all your gateway needs
Declared support is generated from the gateway repository support matrix and conformance profile
AIService and TokenPolicy are disabled by default and configure model providers and token policy when enabled
The Go control plane watches Kubernetes resources; the Rust data plane receives runtime snapshots over gRPC/xDS
WasmPlugin supports URL, ConfigMap, or inline module sources bound to target resources
Admin APIs, Prometheus metrics, Grafana dashboards, and health probe ports are part of the stack
The Helm chart is published at chart.nantian.dev and repository-local Kustomize overlays are available
How Nantian compares to other Gateway API implementations
| Feature | Nantian Gateway | Envoy Gateway | Traefik | NGINX Gateway |
|---|---|---|---|---|
| Gateway API | ✅ v1.5.1 | ✅ | ✅ | ✅ |
| Control Plane | Go | Go | Go | Go |
| Data Plane | Rust 🦀 | C++ | Go | NGINX C |
| AI Gateway | ✅ Experimental | ✅ (AI Gateway) | ❌ | ❌ |
| Wasm Plugins | ✅ Experimental | ✅ | ✅ | ❌ |
| Helm Chart | ✅ | ✅ | ✅ | ✅ |
| License | Apache 2.0 | Apache 2.0 | MIT | Apache 2.0 |
Nantian's key differentiator: Rust data plane for memory safety and high concurrency, plus native AI Gateway capabilities to manage LLM provider traffic without extra components.
From Kubernetes Ingress to AI Gateway — one platform for every scenario
Standard Gateway API compatible, replacing NGINX Ingress. Supports HTTP, gRPC, TCP, UDP, and TLS routing — no custom CRDs required.
Built-in AIService and TokenPolicy resources unify authentication, rate limiting, and routing across OpenAI, Anthropic, Hunyuan, and other model providers.
Inject custom logic into the data plane via WasmPlugin — request rewriting, authentication, observability — hot-loadable without restarting the gateway.
gRPC/xDS protocol distributes routing snapshots. Decoupled control and data planes enable multi-cluster federation with independent scaling.
Add the Helm repository, install the gateway, then follow the full quick start
helm repo add nantian-gw https://chart.nantian.dev
helm repo update
helm install nantian-gw nantian-gw/nantian-gw --namespace nantian-gw --create-namespace