---
title: "MCP Gateway vs MCP Proxy vs AI Gateway vs API Gateway - API7.ai"
description: "Compare MCP gateways, MCP proxies, AI gateways, and API gateways by protocol role, traffic, security, routing, compatibility, and operations."
url: "https://api7.ai/learning-center/api-gateway-guide/ai-gateway-vs-mcp-gateway-vs-api-gateway"
---

[Skip to article](#column-content)

Column contentsAPI Gateway Guide

[API Gateway Guide](https://api7.ai/learning-center/api-gateway-guide)

Search this column

01 Gateway Fundamentals

[View this topic · 10](https://api7.ai/learning-center/api-gateway-guide?topic=fundamentals)[1. What Is an API Gateway? API Gateway Software and Use Cases](https://api7.ai/learning-center/api-gateway-guide/what-is-an-api-gateway)[2. Why Do Modern Architectures Need an API Gateway?](https://api7.ai/learning-center/api-gateway-guide/why-do-modern-architectures-need-api-gateway)[3. Reverse Proxy vs Load Balancer vs API Gateway](https://api7.ai/learning-center/api-gateway-guide/api-gateway-vs-reverse-proxy-vs-load-balancer)[4. API Gateway and Microservices: Decoupling and Traffic Management](https://api7.ai/learning-center/api-gateway-guide/api-gateway-and-microservices)[5. Understanding Core API Gateway Features](https://api7.ai/learning-center/api-gateway-guide/core-api-gateway-features)[6. Open Source vs. Commercial API Gateways: How to Choose the Right One?](https://api7.ai/learning-center/api-gateway-guide/open-source-vs-commercial-api-gateway)[7. API Gateway vs Service Mesh: How to Choose?](https://api7.ai/learning-center/api-gateway-guide/api-gateway-vs-service-mesh)[8. Envoy vs NGINX, Java, and Go for API Gateways](https://api7.ai/learning-center/api-gateway-guide/nginx-vs-envoy-vs-java-vs-go)[9. Core Concepts of API Gateway: Service, Route, Upstream, Plugin, and Consumer](https://api7.ai/learning-center/api-gateway-guide/core-concepts-of-api-gateway)[10. API Gateway Comparison: Apache APISIX vs. Kong vs. Traefik vs. KrakenD vs. Tyk](https://api7.ai/learning-center/api-gateway-guide/api-gateway-comparison-apisix-kong-traefik-krakend-tyk)

02 Gateway Architecture and Design

[View this topic · 10](https://api7.ai/learning-center/api-gateway-guide?topic=architecture)[11. Deep Dive into API Gateway Architecture: Core Components and Design Principles](https://api7.ai/learning-center/api-gateway-guide/api-gateway-architecture)[12. Choosing the Right API Gateway for Your Architecture](https://api7.ai/learning-center/api-gateway-guide/api-gateway-monolithic-vs-microservices)[13. How to Architect an API Gateway for High Availability (HA)?](https://api7.ai/learning-center/api-gateway-guide/api-gateway-high-availability)[14. API Gateway Plugin Architecture: How to Extend and Customize Your API Gateway](https://api7.ai/learning-center/api-gateway-guide/api-gateway-plugin-system)[15. How Event-Driven Architecture (EDA) Works with API Gateway?](https://api7.ai/learning-center/api-gateway-guide/api-gateway-event-driven-architecture)[16. API Gateway Request Lifecycle Explained](https://api7.ai/learning-center/api-gateway-guide/api-gateway-request-lifecycle-explained)[17. Cloud-Native API Gateway on Kubernetes](https://api7.ai/learning-center/api-gateway-guide/cloud-native-api-gateway-kubernetes)[18. API Gateway Control Plane vs. Data Plane](https://api7.ai/learning-center/api-gateway-guide/api-gateway-control-plane-vs-data-plane)[19. How API Gateways Work with GitOps: Automating API Management](https://api7.ai/learning-center/api-gateway-guide/api-gateway-gitops-integration)[20. How to Perform A/B Testing and Canary Releases with an API Gateway](https://api7.ai/learning-center/api-gateway-guide/ab-testing-canary-release-api-gateway)

03 Gateway Setup and Traffic Handling

[View this topic · 10](https://api7.ai/learning-center/api-gateway-guide?topic=traffic-protocols)[21. How to Deploy an API Gateway on Kubernetes: Step-by-Step Guide for Engineers](https://api7.ai/learning-center/api-gateway-guide/deploy-api-gateway-on-kubernetes)[22. Hands-on Setup with Apache APISIX API Gateway](https://api7.ai/learning-center/api-gateway-guide/api-gateway-getting-started-guide)[23. API Gateway Auth: OAuth 2.0, JWT & OIDC with APISIX](https://api7.ai/learning-center/api-gateway-guide/api-gateway-authentication-apache-apisix-oauth-jwt-oidc)[24. Mastering Traffic Control: Efficiently Managing API Requests with Gateways](https://api7.ai/learning-center/api-gateway-guide/api-gateway-traffic-control-strategies)[25. Building Reliable API Gateways with Logging and Monitoring](https://api7.ai/learning-center/api-gateway-guide/api-gateway-logging-monitoring-best-practices)[26. Dynamic Routing in API Gateways: Best Practices for Flexible Traffic Management](https://api7.ai/learning-center/api-gateway-guide/api-gateway-dynamic-routing-management)[27. How API Gateways Support WebSocket](https://api7.ai/learning-center/api-gateway-guide/how-api-gateways-support-websocket)[28. API Gateway for GraphQL: Security, Cost, and Caching](https://api7.ai/learning-center/api-gateway-guide/api-gateway-handle-graphql)[29. How API Gateways Handle HTTP/3 QUIC Requests](https://api7.ai/learning-center/api-gateway-guide/api-gateway-http3-quic)[30. How API Gateways Handle MQTT Requests](https://api7.ai/learning-center/api-gateway-guide/api-gateway-mqtt)

04 Gateway Integrations, AI, and MCP

[View this topic · 10](https://api7.ai/learning-center/api-gateway-guide?topic=ai-mcp)[31. How API Gateways Handle gRPC Requests](https://api7.ai/learning-center/api-gateway-guide/api-gateway-grpc-support)[32. How API Gateways Implement Custom Plugins](https://api7.ai/learning-center/api-gateway-guide/api-gateway-custom-plugins)[33. How API Gateways Integrate with Kafka](https://api7.ai/learning-center/api-gateway-guide/api-gateway-kafka-integration)[34. How to Implement API Aggregation in API Gateways](https://api7.ai/learning-center/api-gateway-guide/api-gateway-api-aggregation)[35. How API Gateways Enable Serverless Architecture](https://api7.ai/learning-center/api-gateway-guide/api-gateway-serverless-architecture)[36. How API Gateways Proxy LLM Requests](https://api7.ai/learning-center/api-gateway-guide/api-gateway-proxy-llm-requests)[37. How API Gateways Enhance MCP Servers](https://api7.ai/learning-center/api-gateway-guide/api-gateway-enhance-mcp-server)[38. MCP Gateway vs MCP Proxy vs AI Gateway vs API Gateway](https://api7.ai/learning-center/api-gateway-guide/ai-gateway-vs-mcp-gateway-vs-api-gateway)[39. What Is an AI Gateway? Architecture and Benefits](https://api7.ai/learning-center/api-gateway-guide/what-is-an-ai-gateway)[40. What Is an MCP Gateway? Architecture, Use Cases & How It Works (2026 Guide)](https://api7.ai/learning-center/api-gateway-guide/what-is-mcp-gateway)

05 Gateway Performance

[View this topic · 10](https://api7.ai/learning-center/api-gateway-guide?topic=performance)[41. How API Gateway Reduces Latency: Performance Optimization Strategies](https://api7.ai/learning-center/api-gateway-guide/how-api-gateway-reduces-latency)[42. API Gateway Caching Strategy: Comprehensive Analysis and Implementation Guide](https://api7.ai/learning-center/api-gateway-guide/api-gateway-caching-strategy-analysis)[43. How to Optimize API Gateway Load Balancing: Advanced Strategies and Best Practices](https://api7.ai/learning-center/api-gateway-guide/how-to-optimize-api-gateway-load-balancing)[44. How API Gateway Efficiently Handles Large-Scale Traffic: Architecture and Strategies](https://api7.ai/learning-center/api-gateway-guide/how-api-gateway-handles-large-scale-traffic)[45. How to Improve API Gateway QoS: Traffic Classes, Capacity Protection, and SLOs](https://api7.ai/learning-center/api-gateway-guide/api-gateway-quality-of-service)[46. API Gateway Log Storage: Pipelines, Retention, Security, and Cost](https://api7.ai/learning-center/api-gateway-guide/api-gateway-log-storage)[47. API Gateway Timeouts and Retries: Safe Policies and APISIX Configuration](https://api7.ai/learning-center/api-gateway-guide/api-gateway-timeouts-retries)[48. API Gateway Concurrency Control: Capacity Budgets, Backpressure, and Load Shedding](https://api7.ai/learning-center/api-gateway-guide/api-gateway-concurrency-control)[49. Asynchronous API Gateway Patterns: Queues, Callbacks, and Performance Boundaries](https://api7.ai/learning-center/api-gateway-guide/asynchronous-api-gateway-patterns)[50. API Gateway TLS Performance: Handshakes, Session Reuse, and Measurement](https://api7.ai/learning-center/api-gateway-guide/api-gateway-tls-performance)

06 Gateway Security

[View this topic · 10](https://api7.ai/learning-center/api-gateway-guide?topic=security)[51. DDoS Defense at the API Gateway: Layered Controls and Failure Planning](https://api7.ai/learning-center/api-gateway-guide/api-gateway-ddos-defense)[52. Distributed API Gateway Rate Limiting: Local vs Redis Counters and Accuracy Trade-offs](https://api7.ai/learning-center/api-gateway-guide/distributed-api-gateway-rate-limiting)[53. API Gateway IP Allowlist and Denylist Management: Trust, Proxies, and Operations](https://api7.ai/learning-center/api-gateway-guide/api-gateway-ip-allowlist-denylist-management)[54. API Gateway Security Scanning: What to Test and How to Operationalize Findings](https://api7.ai/learning-center/api-gateway-guide/api-gateway-security-scanning)[55. SQL Injection and XSS at the API Gateway: Detection Limits and Defense in Depth](https://api7.ai/learning-center/api-gateway-guide/sql-injection-xss-api-gateway)[56. API Access Log Auditing: Evidence, Integrity, Retention, and Review](https://api7.ai/learning-center/api-gateway-guide/api-access-log-auditing)[57. Preventing API Scraping and Abuse: Identity, Behavioral Signals, and Rate Controls](https://api7.ai/learning-center/api-gateway-guide/prevent-api-scraping-abuse)[58. API Gateway mTLS: Client Identity, Certificate Rotation, and APISIX Configuration](https://api7.ai/learning-center/api-gateway-guide/api-gateway-mtls)[59. Fine-Grained API Gateway Authorization: RBAC, ABAC, and External Policy Decisions](https://api7.ai/learning-center/api-gateway-guide/fine-grained-api-gateway-authorization)[60. API Gateway and WAF Reference Architecture: Ownership, Header Trust, and Failure Modes](https://api7.ai/learning-center/api-gateway-guide/api-gateway-waf-reference-architecture)

07 Gateway Ecosystem and Comparisons

[View this topic · 10](https://api7.ai/learning-center/api-gateway-guide?topic=ecosystem)[61. Apache APISIX vs Kong: Migration, Operations, and Team Fit](https://api7.ai/learning-center/api-gateway-guide/apache-apisix-vs-kong)[62. Traefik vs Kong vs APISIX: A Reproducible API Gateway Benchmark Method](https://api7.ai/learning-center/api-gateway-guide/traefik-kong-apisix-benchmark-method)[63. Amazon API Gateway Explained: Deployment Models, Trade-offs, and Use Cases](https://api7.ai/learning-center/api-gateway-guide/amazon-api-gateway-explained)[64. API Gateway and Istio: Clear North-South and East-West Boundaries](https://api7.ai/learning-center/api-gateway-guide/api-gateway-istio)[65. API Gateway with NGINX: Layering, Ownership, and Migration Patterns](https://api7.ai/learning-center/api-gateway-guide/api-gateway-nginx)[66. API Gateway with Envoy: Deployment Patterns and Policy Ownership](https://api7.ai/learning-center/api-gateway-guide/api-gateway-envoy)[67. API Gateway with Cloudflare API Shield: Edge and Origin Security Boundaries](https://api7.ai/learning-center/api-gateway-guide/api-gateway-cloudflare-api-shield)[68. Apigee vs MuleSoft vs Amazon API Gateway: A Decision Framework](https://api7.ai/learning-center/api-gateway-guide/apigee-mulesoft-amazon-api-gateway)[69. Should an API Gateway Connect Directly to a Database? Patterns and Risks](https://api7.ai/learning-center/api-gateway-guide/api-gateway-database-access-patterns)[70. API Gateway Architecture for Large-Scale SaaS Platforms](https://api7.ai/learning-center/api-gateway-guide/api-gateway-large-scale-saas-architecture)

08 Industry Applications and Integrations

[View this topic · 10](https://api7.ai/learning-center/api-gateway-guide?topic=applications)[71. API Gateway CI/CD: Validation, Promotion, and Safe Rollback](https://api7.ai/learning-center/api-gateway-guide/api-gateway-ci-cd)[72. API Gateways for AI and ML Inference: Traffic, Security, and Observability](https://api7.ai/learning-center/api-gateway-guide/api-gateways-ai-ml-inference)[73. API Gateway Architecture for Financial Services: Controls and Failure Boundaries](https://api7.ai/learning-center/api-gateway-guide/api-gateway-financial-services)[74. API Gateway Architecture for E-Commerce: Checkout Reliability and Traffic Peaks](https://api7.ai/learning-center/api-gateway-guide/api-gateway-ecommerce)[75. API Gateways for Mobile Applications: Authentication, Payloads, and Versioning](https://api7.ai/learning-center/api-gateway-guide/api-gateways-mobile-applications)[76. API Gateways for Web3 Applications: RPC Traffic, Identity, and Abuse Controls](https://api7.ai/learning-center/api-gateway-guide/api-gateways-web3-applications)[77. OpenAPI-Driven API Gateway Operations: Import, Validation, and Drift Control](https://api7.ai/learning-center/api-gateway-guide/openapi-driven-api-gateway-operations)[78. gRPC-Web and gRPC Transcoding Through an API Gateway](https://api7.ai/learning-center/api-gateway-guide/grpc-web-grpc-transcoding-api-gateway)[79. API Gateway Pricing and Cost Optimization: A Workload-Based Model](https://api7.ai/learning-center/api-gateway-guide/api-gateway-pricing-cost-optimization)[80. Evaluating API Gateway Directions: Evidence and Architectural Impact](https://api7.ai/learning-center/api-gateway-guide/api-gateway-trends)

09 Advanced Gateway Applications

[View this topic · 10](https://api7.ai/learning-center/api-gateway-guide?topic=operations)[81. API Gateways in Zero Trust Architecture: Identity, Policy, and Limits](https://api7.ai/learning-center/api-gateway-guide/api-gateways-zero-trust-architecture)[82. API Gateway Log Analytics: Correlation, Queries, and Operational Signals](https://api7.ai/learning-center/api-gateway-guide/api-gateway-log-analytics)[83. From API Sprawl to Governed Delivery: The Gateway's Role in Modernization](https://api7.ai/learning-center/api-gateway-guide/api-gateway-modernization)[84. API Monetization at the Gateway: Plans, Metering, Entitlements, and Billing Boundaries](https://api7.ai/learning-center/api-gateway-guide/api-monetization-gateway)[85. Safe AI-Assisted API Gateway Optimization: Recommendations, Guardrails, and Approval](https://api7.ai/learning-center/api-gateway-guide/ai-assisted-api-gateway-optimization)[86. API Gateways for Low-Code and No-Code Integration: Governance and Safety](https://api7.ai/learning-center/api-gateway-guide/api-gateways-low-code-no-code-integration)[87. API Gateway and CIAM Integration: Token Validation, Consent, and Identity Context](https://api7.ai/learning-center/api-gateway-guide/api-gateway-ciam-integration)[88. Elastic API Gateway Scaling: Signals, Capacity, and Safe Scale-In](https://api7.ai/learning-center/api-gateway-guide/elastic-api-gateway-scaling)[89. Multi-Tenant SaaS API Gateways: Isolation, Quotas, and Tenant-Aware Policy](https://api7.ai/learning-center/api-gateway-guide/multi-tenant-saas-api-gateway-policy)[90. API Gateway DevSecOps: Policy as Code, Testing, and Runtime Feedback](https://api7.ai/learning-center/api-gateway-guide/api-gateway-devsecops)

10 Platform Engineering and Planning

[View this topic · 10](https://api7.ai/learning-center/api-gateway-guide?topic=platform-governance)[91. API Discovery and Documentation at the Gateway: Inventory, Ownership, and Drift](https://api7.ai/learning-center/api-gateway-guide/api-discovery-documentation-gateway)[92. WebAssembly Plugins for API Gateways: Portability, Isolation, and APISIX Boundaries](https://api7.ai/learning-center/api-gateway-guide/api-gateway-webassembly-plugins)[93. Monitoring API Gateways with Prometheus and Grafana: Metrics, Dashboards, and Alerts](https://api7.ai/learning-center/api-gateway-guide/monitor-api-gateways-prometheus-grafana)[94. API Gateway Reliability Engineering: SLOs, Failure Isolation, and Incident Response](https://api7.ai/learning-center/api-gateway-guide/api-gateway-reliability-engineering)[95. API Gateways for Edge AI Inference: Routing, Constraints, and Observability](https://api7.ai/learning-center/api-gateway-guide/api-gateways-edge-ai-inference)[96. Reducing Cloud Cost with API Gateway Controls: Measurement Before Optimization](https://api7.ai/learning-center/api-gateway-guide/reducing-cloud-cost-api-gateway-controls)[97. Improving API Developer Experience with Gateway Standards and Self-Service](https://api7.ai/learning-center/api-gateway-guide/improving-api-developer-experience-gateway-self-service)[98. API Gateway Change Management: Compatibility, Progressive Delivery, and Rollback](https://api7.ai/learning-center/api-gateway-guide/api-gateway-change-management-rollback)[99. Operating Serverless APIs: Cold Starts, Idempotency, and Asynchronous Handoffs](https://api7.ai/learning-center/api-gateway-guide/operating-serverless-apis-gateway)[100. Planning an API Gateway Roadmap for the Next Five Years](https://api7.ai/learning-center/api-gateway-guide/planning-api-gateway-roadmap-next-five-years)

[API Gateway Guide](https://api7.ai/learning-center/api-gateway-guide)[Gateway Integrations, AI, and MCP](https://api7.ai/learning-center/api-gateway-guide?topic=ai-mcp)

On this page

On this page

- [Key Takeaways](#key-takeaways)

- [Start with the Standard MCP Roles](#start-with-the-standard-mcp-roles)

- [What Is an API Gateway?](#what-is-an-api-gateway)

- [What Is an AI Gateway?](#what-is-an-ai-gateway)

- [What Is an MCP Proxy?](#what-is-an-mcp-proxy)

- [What Is an MCP Gateway?](#what-is-an-mcp-gateway)

- [Side-by-Side Comparison](#side-by-side-comparison)

- [Current MCP Transport Behavior](#current-mcp-transport-behavior)

  - [Streamable HTTP in 2026-07-28](#streamable-http-in-2026-07-28)
  - [Legacy 2025-Era Compatibility](#legacy-2025-era-compatibility)

- [Authentication and Authorization Boundaries](#authentication-and-authorization-boundaries)

- [Retry, Failure, and Application State](#retry-failure-and-application-state)

- [A Reference Deployment](#a-reference-deployment)

- [Where Apache APISIX and AISIX Fit](#where-apache-apisix-and-aisix-fit)

- [Selection Checklist](#selection-checklist)

- [FAQ](#faq)

  - [Is an MCP gateway part of the MCP standard?](#is-an-mcp-gateway-part-of-the-mcp-standard)
  - [Is an MCP proxy the same as an MCP gateway?](#is-an-mcp-proxy-the-same-as-an-mcp-gateway)
  - [Does current MCP use sessions?](#does-current-mcp-use-sessions)
  - [Does current MCP require SSE?](#does-current-mcp-require-sse)
  - [Can a standard API gateway protect an MCP server?](#can-a-standard-api-gateway-protect-an-mcp-server)
  - [Should a gateway retry MCP tool calls?](#should-a-gateway-retry-mcp-tool-calls)

- [Next Steps](#next-steps)

Article 38

# MCP Gateway vs MCP Proxy vs AI Gateway vs API Gateway

API7.ai

September 28, 2026

An **MCP gateway**, **MCP proxy**, **AI gateway**, and **API gateway** can all appear between a client and an upstream system. The important difference is not where the component is drawn. It is which protocol and policy boundary the component owns.

A standard API gateway governs HTTP, gRPC, or other API traffic. An AI gateway adds controls for model-provider APIs, such as token accounting and model-aware routing. An MCP gateway governs Model Context Protocol requests and tool access. An MCP proxy may only bridge or forward MCP traffic. These are deployment patterns, not four mutually exclusive product categories.

This guide uses the current **MCP 2026-07-28** specification. That revision made the protocol core stateless and removed the previous initialization handshake and protocol-level sessions. A separate compatibility section explains the 2025-era behavior that existing clients and servers may still use.

For the deployment and operating model around model traffic, agents, and MCP tools, use the [AI Gateway Guide](https://api7.ai/learning-center/ai-gateway-guide) as the cluster overview.

## Key Takeaways

- The MCP specification defines hosts, clients, and servers. It does not standardize an “MCP gateway” or “MCP proxy” role.
- MCP 2026-07-28 is stateless at the protocol layer. Each request carries its protocol version and client capabilities; `server/discover` provides optional up-front discovery.
- Current Streamable HTTP uses POST requests to one MCP endpoint. `Mcp-Method` and `Mcp-Name` headers let an intermediary route or authorize without parsing the JSON body.
- MCP authorization is optional. When used over HTTP, access tokens must be validated for the intended MCP server; token passthrough is prohibited.
- A protocol-level retry does not make a state-changing tool idempotent. Tool and application semantics must prevent duplicate side effects.

## Start with the Standard MCP Roles

The current [MCP architecture](https://modelcontextprotocol.io/specification/2026-07-28/architecture) is host-client-server:

- A **host** is the AI application that coordinates one or more MCP clients and enforces user-facing security and consent decisions.
- An **MCP client** is created by the host, communicates with exactly one server, and attaches protocol version and client capabilities to every request.
- An **MCP server** exposes tools, resources, or prompts to that client.

The protocol does not reserve a gateway role. A component marketed as an MCP gateway therefore needs a concrete definition: Does it terminate and originate MCP connections? Does it aggregate tools from several servers? Does it only forward messages? Does it make tool-level authorization decisions? Those answers matter more than the label.

```
flowchart LR
  U[User] --> H[AI Host]
  H --> C1[MCP Client A]
  H --> C2[MCP Client B]
  C1 --> G[MCP Policy Layer]
  C2 --> G
  G --> S1[MCP Server A]
  G --> S2[MCP Server B]
```

MCP 2026-07-28 is stateless at the protocol layer: each request is self-contained. If an application needs state across calls, a server can return an explicit handle that a later tool call supplies as an ordinary argument. This is different from hiding application state in a transport session.

## What Is an API Gateway?

An [API gateway](https://api7.ai/learning-center/api-gateway-guide/what-is-an-api-gateway) is an API-focused reverse proxy and policy enforcement point. Depending on the product and configuration, it may provide:

- routing and upstream load balancing;
- TLS termination and service-to-service TLS;
- authentication and authorization integration;
- [rate limiting](https://api7.ai/blog/token-bucket-vs-leaky-best-rate-limiting-algorithm);
- request and response transformation;
- logs, metrics, and trace propagation; and
- lifecycle controls such as route publication and policy rollout.

Those capabilities apply to normal API traffic. They do not automatically make a gateway aware of MCP methods, tools, capabilities, or compatibility eras. A generic gateway can still provide an outer network and identity boundary for an MCP server, while MCP-specific policy requires MCP-aware processing or the standardized MCP routing headers.

## What Is an AI Gateway?

An **AI gateway** focuses on calls from applications to model-provider APIs. Common responsibilities include:

- routing across approved models or providers;
- adapting provider-specific request formats;
- measuring input and output tokens;
- enforcing model, tenant, or budget policy;
- applying prompt and response controls where appropriate;
- preserving streamed model responses; and
- recording model latency, errors, and usage.

An AI gateway handles the **application-to-model** path. MCP handles the **host-to-tool or context server** path. One user task may cross both: an agent calls a model through an AI gateway, then uses MCP to invoke a tool through an MCP gateway.

## What Is an MCP Proxy?

“MCP proxy” has no single standards definition. In practice, it often means a narrow intermediary that does one or more of the following:

- exposes a remote MCP server through a different network endpoint;
- bridges a local transport to a remote transport;
- forwards Streamable HTTP while applying basic network controls; or
- presents one or more upstream servers behind a stable address.

A proxy may be useful without being a governance platform. Before adopting one, determine which protocol revisions it accepts, whether it translates between modern and legacy eras, how it validates mirrored headers against body metadata, and whether it transforms or only forwards MCP messages.

## What Is an MCP Gateway?

An **MCP gateway** is an architectural pattern for applying centralized policy to agent-to-tool traffic. A production implementation may provide:

- identity-aware access to servers, tools, resources, or prompts;
- tool allowlists scoped to a user, agent, tenant, or environment;
- discovery or aggregation of approved tools;
- MCP request and result audit records;
- rate, concurrency, or budget controls;
- protocol-version and capability checks; and
- routing to eligible, authorized MCP server instances.

MCP 2026-07-28 improves intermediary visibility. Streamable HTTP requests mirror the MCP method and, where applicable, object name into required `Mcp-Method` and `Mcp-Name` headers. The message body remains the source of truth, and the transport defines how a mismatch must be rejected. A gateway can use those headers for routing or policy without turning every JSON body into a custom parsing contract.

An MCP gateway can be built beside a standard API gateway, behind it, or as an MCP-aware capability within the same platform. The architecture should avoid two policy layers making contradictory identity, timeout, or retry decisions.

## Side-by-Side Comparison

| Dimension             | API gateway                | AI gateway                                 | MCP proxy                              | MCP gateway                                                            |
| --------------------- | -------------------------- | ------------------------------------------ | -------------------------------------- | ---------------------------------------------------------------------- |
| Primary traffic       | APIs such as HTTP or gRPC  | Model-provider APIs                        | MCP transport traffic                  | MCP methods and tool traffic                                           |
| Protocol awareness    | API protocol and route     | Model request, response, stream, and usage | Varies; often transport-focused        | MCP version, headers, capabilities, and operations                     |
| Typical policy object | Route, service, consumer   | Model, provider, tenant, token budget      | Endpoint or connection                 | Server, tool, principal, request                                       |
| State responsibility  | Usually request-oriented   | Provider stream or application metadata    | Depends on translation mode            | Modern core is stateless; explicit application handles remain possible |
| Main value            | Central API traffic policy | Model governance and operations            | Connectivity or compatibility bridging | Tool governance and auditability                                       |
| Standardized role     | No; deployment pattern     | No; deployment pattern                     | No                                     | No                                                                     |

The rows describe common patterns, not guaranteed product features. Verify each implementation against the protocol revisions and controls you need.

## Current MCP Transport Behavior

The current [transport specification](https://modelcontextprotocol.io/specification/2026-07-28/basic/transports) defines `stdio` and **Streamable HTTP** as standard transports; custom transports are also allowed when they preserve protocol semantics.

### Streamable HTTP in 2026-07-28

Each message is an HTTP POST to one MCP endpoint. A reply is either a JSON object or a request-scoped SSE stream. Every request carries protocol version and client capabilities in `_meta.io.modelcontextprotocol/*`; Streamable HTTP mirrors selected values into headers for intermediaries.

Important current behavior includes:

- no `initialize`/`notifications/initialized` handshake;
- no protocol-level `Mcp-Session-Id`;
- `server/discover`, which a server must implement and a client may call for supported versions and capabilities;
- required `Mcp-Method` and `Mcp-Name` request headers where defined;
- `subscriptions/listen`, a long-lived POST response stream for opted-in change notifications; and
- no SSE event-ID resumability for a broken response stream.

The [2026-07-28 key changes](https://modelcontextprotocol.io/specification/2026-07-28/changelog) also make list results cacheable using `ttlMs` and `cacheScope`. A shared intermediary must respect private cache scope and must never let one principal receive another principal's tool or resource listing.

### Legacy 2025-Era Compatibility

Earlier revisions through 2025-11-25 used an initialization exchange and could establish a connection-scoped session with `Mcp-Session-Id`. The older Streamable HTTP design allowed an HTTP GET listening stream, and still older HTTP+SSE used separate endpoints. These behaviors are legacy compatibility concerns, not the current architecture.

A gateway supporting both eras should:

- detect and validate the declared protocol version;
- keep 2025-era session routing or shared state isolated from modern stateless traffic;
- avoid inventing a session ID for 2026-07-28 requests;
- translate only when semantics are fully defined; and
- log the negotiated or declared era so incidents can be diagnosed.

HTTP+SSE is deprecated under the current feature-lifecycle policy. New implementations should use current Streamable HTTP.

## Authentication and Authorization Boundaries

MCP authorization is optional. When an HTTP deployment uses the current [MCP authorization framework](https://modelcontextprotocol.io/specification/2026-07-28/basic/authorization), the protected MCP server acts as the OAuth resource server. It must accept only access tokens intended for that server and must not pass the caller's token through to an upstream resource.

A sound gateway design answers four questions:

1. **Who authenticates the principal?** This may be an identity provider integrated with the host or gateway.
2. **Who validates the access token?** Validate issuer, intended resource or audience, time constraints, and permissions using the token format and authorization-server contract.
3. **Who authorizes the specific MCP operation or tool?** URL-level permission is often too broad for a server exposing tools with different effects.
4. **Which identity reaches a downstream system?** Use an explicit delegation or service-identity design. Do not blindly forward the caller's bearer token.

Client and server self-identification in `_meta` is useful for display, logging, and debugging; it is not authenticated identity and must not drive a security decision by itself. Use short-lived credentials, least privilege, and an auditable mapping from the authenticated principal to tool authorization. Treat tool arguments and results as untrusted data.

## Retry, Failure, and Application State

MCP 2026-07-28 removed resumability for a broken Streamable HTTP response. A retried request uses a new request ID. That protocol behavior does not prove that a state-changing tool made no change before the response was lost.

- Give state-changing tools an end-to-end idempotency or deduplication mechanism where replay is possible.
- If the outcome is uncertain and no such mechanism exists, surface that uncertainty instead of silently calling the tool again.
- Pass explicit server-minted handles when an application workflow needs state across calls.
- Bound total deadline and retry count; honor cancellation where the transport supports it.
- Record the original operation identity and every attempt for investigation.

Round-robin routing is straightforward for stateless protocol requests, but application handles may still refer to state stored by a service. Design that storage and failover explicitly.

## A Reference Deployment

```
sequenceDiagram
  participant Host as AI Host / MCP Client
  participant Edge as API Gateway
  participant MCPG as MCP-Aware Gateway
  participant IdP as Identity and Policy Service
  participant Server as MCP Server

  Host->>Edge: Streamable HTTP POST with credential
  Edge->>Edge: TLS, token validation, network policy
  Edge->>MCPG: Authenticated principal and protected context
  MCPG->>MCPG: Validate MCP version and mirrored headers
  MCPG->>IdP: Authorize principal, server, and tool
  IdP-->>MCPG: Scoped decision
  MCPG->>Server: Self-contained MCP request
  Server-->>MCPG: JSON or request-scoped stream
  MCPG-->>Host: MCP response
```

The edge and MCP-aware layers may be separate services or capabilities of one platform. If separate, the gateway must strip or overwrite untrusted client-supplied identity headers, protect the gateway-to-MCP hop, and let the server trust identity context only from an authenticated gateway path.

## Where Apache APISIX and AISIX Fit

[Apache APISIX](https://api7.ai/apisix) can provide an API traffic layer with routing, authentication integrations, rate limits, observability, and streaming proxy behavior when configured for the workload. Its [AI Proxy plugin](https://apisix.apache.org/docs/apisix/plugins/ai-proxy/) supports model-provider traffic, but a standard APISIX route should not be described as an MCP-aware gateway by default.

[AISIX](https://api7.ai/ai-gateway) provides MCP gateway capabilities for publishing and governing MCP tools and supports the current MCP 2026-07-28 era. Verify the documented protocol mode, tool-access model, and deployment topology for the release you operate. Separating these claims matters: APISIX is the open-source API gateway foundation, while MCP aggregation and tool governance require explicit product support.

## Selection Checklist

Choose the smallest architecture that meets the policy boundary:

- Use a **standard API gateway** for normal API routing, identity enforcement, traffic policy, and observability.
- Add an **AI gateway** for model-provider routing, token usage, model policy, or AI-specific telemetry.
- Use an **MCP proxy** for a narrow, documented connectivity or protocol-compatibility problem.
- Use an **MCP gateway** when agents or users need centralized, tool-level authorization, discovery, quotas, and audit controls.

Before production, test:

- 2026-07-28 request metadata and `server/discover`;
- `Mcp-Method`/`Mcp-Name` mismatch rejection;
- JSON and request-scoped streaming responses;
- `subscriptions/listen` authorization and cancellation;
- expired, mis-audienced, and under-scoped tokens;
- unauthorized or unknown tools;
- uncertain outcomes from state-changing calls; and
- legacy-era isolation if 2025 clients are supported.

## FAQ

### Is an MCP gateway part of the MCP standard?

No. MCP standardizes host, client, server, messages, transports, and optional authorization. “MCP gateway” is a deployment pattern for applying policy and routing between clients and servers.

### Is an MCP proxy the same as an MCP gateway?

Not necessarily. A proxy may only forward or bridge transports. A gateway usually implies centralized identity, authorization, tool policy, audit, and operational controls. Product names alone do not prove those capabilities.

### Does current MCP use sessions?

MCP 2026-07-28 removed the protocol-level session and `Mcp-Session-Id`. Applications can still use explicit state handles passed as tool arguments. Clients and servers supporting 2025-era revisions may retain legacy session behavior for compatibility.

### Does current MCP require SSE?

Each Streamable HTTP message is a POST. A response can be a JSON object or a request-scoped SSE stream, and notification subscriptions use a long-lived `subscriptions/listen` POST response. The old two-endpoint HTTP+SSE transport is deprecated.

### Can a standard API gateway protect an MCP server?

Yes, it can provide TLS, token validation, network controls, rate limits, and observability. Tool-level authorization and protocol-version handling require MCP-aware policy or use of the standardized MCP headers.

### Should a gateway retry MCP tool calls?

Only when the operation is idempotent or protected by an end-to-end deduplication mechanism. A lost response does not prove that the tool made no change.

## Next Steps

- Review the current [MCP 2026-07-28 specification](https://modelcontextprotocol.io/specification/2026-07-28).
- Follow [Set Up MCP Gateway](https://docs.api7.ai/ai-gateway/mcp-gateway/setup) for current AISIX behavior.
- Review [What Is an API Gateway?](https://api7.ai/learning-center/api-gateway-guide/what-is-an-api-gateway).
- Compare [API gateway platforms](https://api7.ai/learning-center/api-gateway-guide/api-gateway-comparison-apisix-kong-traefik-krakend-tyk).
- Explore [AISIX AI Gateway](https://api7.ai/ai-gateway) for current AI and MCP gateway capabilities.

[Previous article**How API Gateways Enhance MCP Servers**](https://api7.ai/learning-center/api-gateway-guide/api-gateway-enhance-mcp-server)[Next article**What Is an AI Gateway? Architecture and Benefits**](https://api7.ai/learning-center/api-gateway-guide/what-is-an-ai-gateway)

On this page

- [Key Takeaways](#key-takeaways)

- [Start with the Standard MCP Roles](#start-with-the-standard-mcp-roles)

- [What Is an API Gateway?](#what-is-an-api-gateway)

- [What Is an AI Gateway?](#what-is-an-ai-gateway)

- [What Is an MCP Proxy?](#what-is-an-mcp-proxy)

- [What Is an MCP Gateway?](#what-is-an-mcp-gateway)

- [Side-by-Side Comparison](#side-by-side-comparison)

- [Current MCP Transport Behavior](#current-mcp-transport-behavior)

  - [Streamable HTTP in 2026-07-28](#streamable-http-in-2026-07-28)
  - [Legacy 2025-Era Compatibility](#legacy-2025-era-compatibility)

- [Authentication and Authorization Boundaries](#authentication-and-authorization-boundaries)

- [Retry, Failure, and Application State](#retry-failure-and-application-state)

- [A Reference Deployment](#a-reference-deployment)

- [Where Apache APISIX and AISIX Fit](#where-apache-apisix-and-aisix-fit)

- [Selection Checklist](#selection-checklist)

- [FAQ](#faq)

  - [Is an MCP gateway part of the MCP standard?](#is-an-mcp-gateway-part-of-the-mcp-standard)
  - [Is an MCP proxy the same as an MCP gateway?](#is-an-mcp-proxy-the-same-as-an-mcp-gateway)
  - [Does current MCP use sessions?](#does-current-mcp-use-sessions)
  - [Does current MCP require SSE?](#does-current-mcp-require-sse)
  - [Can a standard API gateway protect an MCP server?](#can-a-standard-api-gateway-protect-an-mcp-server)
  - [Should a gateway retry MCP tool calls?](#should-a-gateway-retry-mcp-tool-calls)

- [Next Steps](#next-steps)
