What AI said about AI gateways in September 2026

Short answer: the AI gateways people actually shortlist in 2026

If I had to pick top brands/products worth evaluating, I’d start with these:

My recommendations by situation

Pick LiteLLM if you want the most proven open-source/self-hosted option and your team can operate infrastructure.

Pick Portkey if you want the most complete managed production gateway: routing, retries, fallbacks, budgets, guardrails, and governance in one place.

Pick OpenRouter if you mainly want access to many models quickly with one API/key and you care more about model breadth than enterprise controls.

Pick Helicone if your #1 problem is LLM observability, cost tracking, request logs, and debugging.

Pick Cloudflare AI Gateway if you already use Cloudflare and want a low-friction, low-cost edge gateway.

Pick Kong AI Gateway if your company already uses Kong or needs API-management-grade enterprise controls.

Pick Vercel AI Gateway if your app is built on Vercel + AI SDK.

Pick Azure API Management AI Gateway if you are an Azure enterprise and want GenAI governance integrated with existing APIM patterns.

If you want a default shortlist

For most teams, I’d evaluate only these four first:

1. LiteLLM — best OSS/self-hosted baseline.
2. Portkey — best managed production/governance gateway.
3. OpenRouter — best model marketplace / quick multi-model access.
4. Helicone — best observability-first gateway.

Then add Cloudflare, Kong, Vercel, or Azure APIM only if you’re already in that ecosystem.

Based on current market coverage, here's a rundown of the AI gateways that come up most often across comparisons and appear to have the widest actual adoption:

Most widely used / recommended

LiteLLM — By far the most frequently cited as the default choice for developers. It is one of the most popular open-source AI gateways, beloved for wrapping 100+ providers behind a single OpenAI-compatible endpoint with minimal friction. It is MIT-licensed at the core, has a large community, and processes enormous request volumes, making it a default choice for developers who want provider unification fast. That said, one comparison notes it suffers from high latency, struggles beyond moderate RPS, lacks built-in scaling, and is best for light or prototype workloads compared to some commercial alternatives.

Portkey — A major enterprise player, recently acquired. Portkey is a widely adopted LLM gateway and observability platform positioned as a control plane for AI, with guardrails, PII redaction, and audit trails built into the gateway layer. It routes to a very large catalog of models (it cites 1,600+ models across 40+ providers), and its gateway core was open-sourced under Apache 2.0. Notably, Portkey was acquired by Palo Alto Networks, with the deal closing in May 2026, and it now serves as the AI Gateway for Prisma AIRS — a strong signal of enterprise validation, though it's worth asking how contracts and roadmaps shift post-acquisition.

Cloudflare AI Gateway — Popular for teams already on Cloudflare's edge network. Cloudflare AI Gateway is the best low-cost choice when Cloudflare already owns the application's edge, security, or Workers runtime. Core analytics, caching, and rate limiting are free on every plan.

OpenRouter — Popular with individual developers and smaller teams for its huge model catalog, and now a huge enterprise validation signal: Axios reported on 17 August 2026 that Stripe had agreed to acquire OpenRouter for more than $8 billion in cash and stock, though the companies had not commented and an official announcement was still expected.

Kong AI Gateway — Comes from Kong's established API gateway background, frequently listed alongside Portkey and LiteLLM in enterprise comparisons as a mature infrastructure option.

Vercel AI Gateway — Called out as the best managed default for multi-model apps, especially for teams already in the Vercel/Next.js ecosystem.

Enterprise/production-focused options

TrueFoundry — Emphasizes raw performance: TrueFoundry AI Gateway delivers ~3–4 ms latency, handles 350+ RPS on 1 vCPU, scales horizontally with ease, and is production-ready.

Bifrost (by Maxim AI) — A newer, performance-focused entrant claiming 11 microseconds of overhead per request at 5,000 RPS, delivering 50× faster performance than Python-based alternatives, which matters critically in production environments serving millions of requests daily.

Helicone — Frequently listed alongside Portkey/LiteLLM/Kong in comparison guides, generally focused on observability plus gateway functionality.

How to pick

The realistic guidance across most of these sources agrees on one point: choosing an AI gateway is about finding the use cases that are most applicable to your organization — it depends on your team's AI maturity, model diversity/lock-in risk

As organizations transition from basic LLM integration to production-grade AI agents and multi-model systems, directly calling LLM provider APIs quickly becomes a bottleneck. Managing multiple provider APIs, rate limits, latency, cost tracking, security guardrails, and key access requires an AI Gateway (or LLM Gateway) to act as a centralized control plane.

A highly-rated selection of the top AI gateways—categorized by what teams actually use them for—provides options ranging from lightweight open-source proxies to heavy-duty enterprise security layers:

1. LiteLLM

• The Category: The Open-Source & Developer Favorite (MIT Licensed).
• Best For: Teams that want to self-host and need the broadest possible model integration.
• How it works: LiteLLM is a proxy that acts as a translator. You point your code to a single, local OpenAI-compatible endpoint, and LiteLLM translates those calls to over 100+ different LLM providers (Anthropic, Bedrock, Azure, Hugging Face, Cohere, etc.).
• Key Strengths:
  • Massive provider coverage (100+ models).
  • Comprehensive virtual key management, user budget limits, and token-based rate limiting.
  • Highly customizable python SDK/proxy.
• Drawbacks: Written in Python, which introduces higher latency overhead (~8ms) compared to newer Go-built gateways. Some developers report that at massive scale, the rapidly updated codebase can occasionally feel buggy.

2. Portkey (Acquired by Palo Alto Networks)

• The Category: The Observability & Security Standard.
• Best For: Growth-to-enterprise level teams prioritizing rigorous compliance, observability, and agentic security.
• How it works: Originally a developer-focused SaaS/hybrid platform, Portkey was acquired by Palo Alto Networks to serve as the core AI gateway for their Prisma AIRS platform. It focuses on securing, monitoring, and tracing the complex data flows of autonomous AI agents.
• Key Strengths:
  • Best-in-class logging, deep analytics, and prompt playground tools.
  • High-grade enterprise security, prompt filtering, PII masking, and threat mitigation built-in.
  • Excellent routing rules, automatic retries, and fallback logic.
• Drawbacks: Might be considered "overkill" if you only need a simple, lightweight logging proxy.

3. Bifrost (by Maxim AI)

• The Category: The Ultra-Low Latency & Agent-First Gateway.
• Best For: High-throughput, real-time applications and agentic workflows where every millisecond matters.
• How it works: An Apache 2.0 open-source gateway built entirely in Go. Bifrost is designed to sit inside your VPC to minimize networking hop costs.
• Key Strengths:
  • Near-zero latency overhead (measured at roughly ~11 microseconds, compared to milliseconds in other gateways).
  • Native support for Model Context Protocol (MCP), allowing AI agents to dynamically interact with external databases and tools.
  • Built-in semantic caching (using vector DBs like Weaviate) to drastically lower repetitive API costs.
• Drawbacks: Fewer model provider integrations natively supported (around 20+) compared to LiteLLM.

4. Cloudflare AI Gateway

• The Category: The Zero-Ops Edge Gateway.
• Best For: Developers and startups that want to start scaling instantly without managing any database or server infrastructure.
• How it works: Sits on Cloudflare's global edge network. You simply swap out your base LLM API URLs with a Cloudflare-generated routing URL.
• Key Strengths:
  • Truly zero-config; you can deploy in minutes.
  • Excellent free tier that covers caching, basic analytics, rate limiting, and request logging.
  • Negligible latency overhead since routing is handled directly at Cloudflare's edge.
• Drawbacks: Highly tied to Cloudflare's cloud ecosystem (not a good fit for on-premises/air-gapped networks). Custom routing rules are not as flexible as code-based self-hosted options.

5. Kong AI Gateway

• The Category: The Infrastructure-First Giant.
• Best For: Established enterprises that already run their services on traditional API gateways and want a unified microservices stack.
• How it works: Kong is one of the world's most battle-tested API gateways. By deploying their specialized AI Gateway plugins, you can route, secure, and monitor LLM traffic alongside your standard REST/GraphQL microservices traffic.
• Key Strengths:
  • Phenomenal, enterprise-grade scalability and robust token-based rate limiting.
  • Seamlessly integrates into existing Kubernetes and enterprise network topologies.
  • No need to adopt a "new" point tool if your team is already a Kong shop.
• Drawbacks: Has a heavy learning curve and configuration overhead. It is not a native "LLM-first" platform, meaning developer setups like prompt versioning and evaluation take more effort to configure.

6. OpenRouter

• The Category: The Ultimate SaaS Model Router.
• Best For: Prototyping, testing, and apps that frequently swap back and forth between open-source models and proprietary models.
• How it works: OpenRouter acts as a managed routing hub. Rather than signing up for 10 different platforms (Anthropic, DeepSeek, OpenAI, Mistral, etc.), you fund one OpenRouter account and use their unified API endpoint to ping any model you want.
• Key Strengths:
  • Offers access to hundreds of open-source and specialized models.
  • Single, consolidated billing—saving you from managing dozens of provider cards.
• Drawbacks: Because it is a fully managed third-party platform, it is not suited for strictly regulated or enterprise industries that require data-sovereign VPC boundaries.

Which one should you choose?

• Choose LiteLLM if you are an independent developer or a growing team that wants a free, highly extensible, and community-loved tool to orchestrate and budget over dozens of model APIs.
• Choose Bifrost if you are building complex autonomous AI agents and cannot afford any latency overhead in your RAG or tool-calling loops.
• Choose Portkey (via Prisma AIRS) if you are in a regulated industry (finance, healthcare, cybersecurity) and require strict data auditing, PII scrubbing, and agentic identity controls.
• Choose Cloudflare AI Gateway if you want the fastest, easiest setup to get analytics, basic caching, and rate-limiting working today.

The best AI gateways people actually use today are usually LiteLLM, Kong AI Gateway, Cloudflare AI Gateway, Portkey, Helicone, OpenRouter, and Vercel AI Gateway. If you want the safest “default” picks by use case, LiteLLM is the leading self-hosted/open-source option, Kong is a strong enterprise platform choice, Cloudflare is the easiest edge-based gateway, and Portkey is one of the most common managed routing/guardrails products.

If you want the shortest recommendation list:

• Best overall open-source: LiteLLM.
• Best enterprise API-platform gateway: Kong AI Gateway.
• Best edge gateway: Cloudflare AI Gateway.
• Best managed production router: Portkey.
• Best observability-focused choice: Helicone.
• Best for widest model choice: OpenRouter.

A few newer enterprise-oriented products also show up repeatedly in 2026 comparisons, especially nexos.ai, Braintrust Gateway, TrueFoundry, Opper, SS&C AI Gateway, and Bifrost, but the most broadly “used in practice” names across multiple comparisons are still the ones above.

If you want, I can narrow this to:

• best for startups
• best open-source
• best enterprise/compliance
• best cheapest option
• best for OpenAI-compatible routing

There’s no single “best” AI gateway — it depends on whether you want a hosted SaaS, a self-hosted open-source proxy, or a cloud-provider-native gateway. But these are the ones teams actually use in production:

🔥 Most Popular AI Gateways (Real-World Usage)

1. LiteLLM — Best open-source / self-hosted

• What it is: A lightweight Python proxy that exposes 100+ LLMs (OpenAI, Anthropic, Google, Bedrock, Azure, etc.) behind one unified API format (OpenAI-compatible).
• Why people use it: Simple to deploy (Docker), supports fallbacks, load balancing, cost tracking, and key management. Great for startups and enterprises that want control.
• Use case: Teams that want to avoid vendor lock-in and standardize on one SDK.

2. Portkey.ai — Best full-featured SaaS gateway

• What it is: AI Gateway + Observability + Guardrails + Prompt management in one platform. Acts as a reverse proxy to 250+ models.
• Why people use it: Excellent dashboard for logs, tokens, cost, latency; supports caching, retries, routing (e.g., cheapest model), and team API keys.
• Use case: Production apps that need observability and governance without building it in-house.

3. Cloudflare AI Gateway — Best for global edge performance

• What it is: Cloudflare’s managed gateway that sits in front of AI APIs and gives you caching, rate limiting, logging, and analytics at the edge.
• Why people use it: Works with Cloudflare Workers easily, adds low latency, and can fall back across providers. Free tier available.
• Use case: Serverless/edge apps, or teams already on Cloudflare.

4. OpenRouter — Best for hobbyists / developer exploration

• What it is: A public marketplace and unified gateway to almost every LLM (open-source and proprietary). One API key to access all models.
• Why people use it: Simple, pay-as-you-go, no infrastructure, includes community-ranked models. Great for prototyping and comparing models.
• Use case: Developers who want to test many models quickly with one integration.

5. AWS/Azure/GCP Native Gateways — Best if you’re deeply cloud-locked

• Amazon Bedrock: One API for FM models (Claude, Llama, Titan, Mistral) with AWS IAM security, VPC integration, and enterprise guardrails.
• Azure API Management (APIM) + Azure OpenAI Service: Very popular in enterprise. APIM acts as the gateway with rate limits, OAuth, and logging in front of Azure OpenAI.
• Google Cloud Vertex AI: Unified API for Gemini, Claude, Llama, and more, with enterprise controls and Vertex AI Gateway features.
• Use case: Enterprises that already run on a cloud provider and prioritize security/compliance.

6. Kong AI Gateway — Best for experienced API teams

• What it is: The AI plugin/extensions built into Kong Gateway (open-source/Hybrid). Supports routing, prompt engineering, LLM access control, and observability.
• Why people use it: If you already use Kong for traditional APIs, it extends the same controls (auth, rate limiting, plugins) to AI APIs.
• Use case: Organizations with mature API management that want LLM APIs to live alongside existing services.

📊 Quick Comparison

✅ How to choose

• Want open-source and self-hosted? → LiteLLM
• Want a polished managed dashboard? → Portkey
• Want global low-latency caching? → Cloudflare AI Gateway
• Want to quickly test models? → OpenRouter
• Already deep in AWS/Azure/GCP? → Use their native gateway
• Need enterprise API management? → Kong

Most production teams combine LiteLLM (as a proxy) with Portkey or Cloudflare (for observability/caching) — or just use Portkey for everything if they don’t mind the SaaS dependency.

LiteLLM, OpenRouter, and Portkey are the most frequently recommended and widely used AI/LLM gateways (also called LLM proxies, routers, or API gateways). These tools sit between your application and multiple AI model providers (OpenAI, Anthropic, Groq, etc.), providing a unified OpenAI-compatible API endpoint while adding routing, fallbacks, cost tracking, caching, observability, guardrails, and more.

They address common pain points like managing multiple provider keys/integrations, handling rate limits or outages via automatic fallbacks, tracking spend across teams/models, and (in some cases) adding semantic caching or PII redaction without changing much application code.

Top Recommendations (Based on 2026 Comparisons and Adoption)

Here are the leaders, drawn from multiple expert roundups, comparisons, and community discussions:

• LiteLLM (Best overall self-hosted/open-source option): MIT-licensed proxy supporting 100+ providers behind one OpenAI-compatible endpoint. Features include virtual keys with budgets/rate limits, automatic fallbacks/load balancing, cost tracking, admin UI/dashboard, and caching options (exact + semantic via Redis/etc.). It is the default for teams wanting full control, zero markup on inference (you use your own keys), and data residency in your VPC/infra. Widely adopted with the largest open-source community and GitHub presence among peers. Enterprise tiers add SSO/RBAC/managed options. Ideal if you have (or are willing to run) some DevOps capacity.

• OpenRouter (Best for zero-ops and broadest model access): Managed SaaS aggregator providing instant access to 300–400+ models from 60–70+ providers via a single API key and endpoint. Pass-through provider pricing (no inference markup) plus a ~5.5% fee on credit purchases. Strong for rapid prototyping, experimentation across models (including niche/open-source ones), consolidated billing, and community-driven routing/metadata. Largest ecosystem/developer community among managed options. Not self-hostable (traffic routes through their platform). Great starting point for small teams or when you want maximum model variety without managing infrastructure.

• Portkey (Best for production observability, guardrails, and governance): Combines gateway functionality with strong built-in features like semantic/exact caching, PII redaction, jailbreak/content filtering, prompt management, audit logs, and deep observability. Open-source gateway core (Apache 2.0) with managed SaaS options (free tier limited logs/requests; paid from ~$49/mo for higher limits/retention). Supports hundreds to 1,600+ models/providers depending on the source. Popular for compliance-focused or team/production workloads. Self-host or VPC options available on higher tiers.

Other Notable Options People Use

• Helicone: Strong observability and logging focus (with routing/caching). Open-source elements; free tier available; paid from ~$79/mo. Good complement or alternative when monitoring is the priority.
• Cloudflare AI Gateway: Free core features (with fees on unified billing/credits), edge caching, analytics, and rate limiting. Excellent if your stack is already on Cloudflare (minimal added latency/infra).
• Vercel AI Gateway: Zero markup, no-retention options, and seamless integration for Next.js/Vercel users.
• Kong AI Gateway: Enterprise-grade if you already run Kong for API management (add AI-specific plugins for proxying, guarding, etc.).

Other mentions in comparisons include Bifrost (high-throughput/low-latency self-hosted), TrueFoundry (enterprise/air-gapped/VPC), and various niche or ecosystem-specific tools.

Quick Decision Guide

• Prototyping, broad models, minimal setup: Start with OpenRouter.
• Self-hosting, control, no markup, own infrastructure: LiteLLM (the community standard).
• Production scale, compliance, guardrails, observability: Portkey (or Helicone for lighter obs focus).
• Already in a specific ecosystem (Cloudflare, Vercel, Kong): Use the native option.
• Many teams combine them (e.g., OpenRouter for experimentation → LiteLLM in production → Portkey/Helicone for logging).

LiteLLM and OpenRouter appear most often as the practical defaults people actually deploy, with Portkey rising for teams needing enterprise features.

The space evolves quickly (acquisitions, open-sourcing moves, new features like advanced caching/routing), so check current pricing, model coverage, and docs for your specific needs (e.g., latency overhead, data residency, or exact provider support). GitHub stars, adoption mentions, and hands-on comparisons consistently highlight the above as the ones in real use.

Top AI gateways for enterprise and advanced AI workloads in 2026 include TrueFoundry, Kong AI Gateway, Portkey, LiteLLM, and Lunar.dev, each with unique strengths in performance, observability, and governance.

1. TrueFoundry AI Gateway

Overview: TrueFoundry is a purpose-built platform designed for enterprises managing multiple AI providers and production-scale workloads. It supports routing across hundreds of LLMs, including OpenAI, Anthropic, Gemini, and self-hosted models TrueFoundry.

Key Strengths:

Sub-5ms latency overhead and high throughput (350+ requests/sec per CPU core) Token-level cost tracking and real-time budget enforcement Strong observability and governance, including SOC 2 Type 2 and HIPAA compliance Supports enterprise features like MCP Gateway, OAuth 2.0, and air-gapped deployments Containerization with GPU optimization and automatic model caching

Best For: Enterprises needing comprehensive governance, cost management, and high-performance AI operations TrueFoundry.

2. Kong AI Gateway

Overview: Kong is an established API management platform that includes AI gateway capabilities for routing and managing LLM traffic TrueFoundry.

Key Strengths:

Integrates with existing Kong ecosystems for organizations already using it Standard enterprise security features (OAuth 2.0, JWT, RBAC) Primarily suitable for enterprise API gateway deployments

Limitations: Lack of LLM-native governance features, limited token-level cost tracking, and layered architecture not optimized for high-volume AI workloads. Best when AI is not the core workload TrueFoundry.

3. Portkey

Overview: Portkey is an LLMOps platform that combines AI gateway functionality with lifecycle management features TrueFoundry.

Key Strengths:

Unified API for hundreds of LLMs with prompt management, guardrails, and 50+ pre-built compliance/security measures Real-time monitoring with SOC2, HIPAA, ISO27001, and GDPR compliance Deployment options include SaaS, hybrid, and fully air-gapped environments 99.99% uptime SLA ensures reliability

Best For: Organizations requiring integrated LLMOps capabilities alongside gateway functionality TrueFoundry.

4. LiteLLM

Overview: Open-source AI gateway for teams emphasizing cost control and flexibility TrueFoundry.

Key Strengths:

Universal API compatibility with support for major LLM providers Docker-based and self-hosted deployments Flexible configuration through YAML and infrastructure-as-code support

Limitations: Lacks enterprise-level support, advanced observability, and robust security features. Some instability at scale with potential latency overhead TrueFoundry.

5. Lunar.dev

Overview: AI gateway combining intelligent routing, caching, rate limiting, and governance for multiple LLM providers TrueFoundry.

Key Strengths:

Unified control plane for AI + MCP + API workloads Implements per-user and per-agent policies for security and auditing Self-hosted VPC, air-gapped deployment options, SOC 2 compliance Granular traffic management including priority queues and prompt transformation

Best For: Security, IT, and platform teams needing centralized governance across AI model traffic TrueFoundry.

Considerations When Choosing an AI Gateway

Performance: TrueFoundry leads in low-latency deployments; LiteLLM may introduce overhead at scale Security & Compliance: Portkey, TrueFoundry, and Lunar.dev offer enterprise-grade compliance; open-source alternatives may lack formal security guarantees Deployment Flexibility: Cloud, hybrid, air-gapped, and VPC deployments vary among providers Feature Requirements: Some gateways focus solely on routing, while others integrate observability, cost tracking, and LLMOps management Use Case Fit: Enterprises with multiple AI providers benefit from TrueFoundry; cost-conscious engineering teams may prefer LiteLLM; existing Kong users may leverage Kong AI Gateway with limitations 2

Overall, TrueFoundry, Portkey, and Lunar.dev are top choices for production-scale enterprise AI, while LiteLLM is ideal for open-source flexibility and cost-conscious teams. Kong AI Gateway serves best as a secondary addition to existing API management infrastructure.

An AI gateway (or LLM gateway) sits as a crucial architectural layer between your applications and multiple AI model providers, standardizing API requests, providing fallback routing, caching responses, and tracking API costs.

The top AI gateways developers and enterprise teams actually use are categorized by their primary deployment style and infrastructure focus:

Open-Source & Self-Hosted Gateways

LiteLLM: Best For: Engineering teams wanting a self-hosted, lightweight option. Compatibility: Translates 100+ LLM providers into OpenAI formats. Trade-off: Python runtime can hit throughput ceilings at scale. Source: Verified as a community favorite on GitHub. Bifrost: Best For: High-performance enterprise production workloads needing sub-millisecond overhead. Architecture: Written in Go with native worker pool isolation. Features: Built-in Model Context Protocol (MCP) routing. Source: Maintained by the team at Maxim AI.

Managed & Enterprise AI Platforms

Portkey: Best For: Teams needing full-stack LLMOps and advanced guardrails. Features: Provides deep prompt management and custom routing. Ecosystem: Backed by Palo Alto Networks' Prisma AIRS portfolio. Source: Evaluated heavily by production security teams via Portkey. Cloudflare AI Gateway: Best For: Teams already building apps inside the Cloudflare ecosystem. Cost: Highly attractive free tier for core features. Edge Native: Handles caching and rate-limiting at the edge. Source: Available natively directly inside the Cloudflare Dashboard.

API-First & Infrastructure Extensions

Kong AI Gateway: Best For: Large organizations with pre-existing Kong API meshes. Infrastructure: Extends the ubiquitous, battle-tested Kong Gateway. Features: Adds AI traffic governance plugins to existing infrastructure. Source: Managed through the enterprise control plane at Kong. OpenRouter: Best For: Rapid prototyping and single-endpoint model aggregation. Catalog: Accesses over 300+ public and open models. Billing: Simplifies multi-provider overhead with consolidated token billing. Source: Widely favored by individual developers using OpenRouter.

Core Selection Framework

To help narrow down the right choice, tell me:

What specific model providers (e.g., OpenAI, Anthropic, local models) do you plan to use? Do you prefer a fully managed SaaS or a self-hosted/open-source deployment inside your own VPC? What is your expected traffic volume (requests per second)?