Kong AI Gateway vs Portkey: Which AI Gateway in 2026?
By API7.ai Team
Last updated: June 2026
Kong AI Gateway and Portkey take two different routes to the same goal: putting one governed API in front of many LLM providers. This guide compares them on architecture, routing, caching, guardrails, observability, governance, self-hosting, and pricing so you can choose the right fit.
Kong AI Gateway adds AI plugins to a mature general-purpose API gateway (self-host, hybrid, or Konnect SaaS), but gates most advanced AI — multi-model load balancing, semantic routing and caching, token/cost limits, semantic/PII guardrails, MCP, and LLM-as-judge — behind Enterprise. Portkey is an AI-native gateway with a hosted governance and observability platform and 1,600+ models, keeping more AI features close to the core, though semantic caching and SSO/SCIM/VPC are Enterprise. The core axis is an API-gateway platform with add-on AI versus an AI-native gateway with hosted governance.
- Teams standardizing on a general-purpose API gateway: Kong AI Gateway
- Teams wanting hosted AI-native governance + observability: Portkey
Kong AI Gateway vs Portkey at a glance
Kong layers AI plugins onto a general-purpose API gateway and gates advanced AI behind Enterprise; Portkey is an AI-native gateway with a hosted governance UI and 1,600+ models. Kong has semantic routing (Enterprise); Portkey's routing is rule-based.
| Dimension | Kong | Portkey |
|---|---|---|
| Best for | AI plugins on a general-purpose API gateway | Hosted AI-native governance + observability |
| Core & runtime | Lua on OpenResty; AI plugins on Kong Gateway | TypeScript gateway + hosted control plane |
| Open-source license | Apache-2.0 core; advanced AI Enterprise | MIT gateway; platform commercial |
| Provider coverage | 15+ providers | 1,600+ models / 45+ providers |
| Semantic routing | ✓ Enterprise (AI Proxy Advanced) | — Rule-based only |
| Ensemble / fusion | — Not documented | — Not documented |
| Caching | Semantic cache (Enterprise) | ✓ Simple + semantic |
| MCP gateway | AI MCP Proxy (Enterprise) | ✓ Documented |
| Self-host / VPC | Docker/K8s, hybrid, Konnect SaaS | OSS self-host; VPC = Enterprise |
| SSO / SCIM | Enterprise/Konnect | Enterprise-only |
What is Kong AI Gateway?
Kong AI Gateway is a set of AI plugins on Kong Gateway, a mature general-purpose API gateway (Apache-2.0 core, Lua/OpenResty), running self-hosted, hybrid, or via the managed Konnect SaaS.
Kong AI Gateway is a set of AI plugins layered on Kong Gateway, a mature general-purpose API gateway (Apache-2.0 core, Lua on OpenResty). It routes to 15+ LLM providers and can run self-hosted, hybrid, or through the managed Konnect SaaS control plane.
Language
Lua (OpenResty)
License
Apache-2.0 core + Enterprise AI plugins
Form factor
API gateway + AI plugins; Konnect SaaS
Best for
AI plugins on a general-purpose gateway
Pros
- AI plugins on a mature, general-purpose API-gateway platform
- Self-host, hybrid, or managed Konnect SaaS control plane
- Semantic routing, semantic cache, and MCP proxy (Enterprise)
- 15+ LLM providers behind one API
Cons
- Free AI surface is thin: OSS AI Proxy handles a single provider/model
- Multi-model LB, semantic routing/cache, token/cost limits are Enterprise
- Semantic/PII guardrails, MCP, LLM-as-judge, SSO/RBAC are Enterprise
What is Portkey?
Portkey is an AI-native gateway pairing an open-source TypeScript gateway with a paid hosted control plane for observability, prompt management, and governance across 1,600+ models.
Portkey is an AI-native gateway that pairs an open-source TypeScript gateway (MIT) with a hosted control plane for observability, prompt management, and governance. It advertises 1,600+ models across 45+ providers.
Language
TypeScript (Node.js)
License
MIT gateway + commercial platform
Form factor
OSS gateway + hosted SaaS
Best for
Hosted observability & governance
Pros
- 1,600+ models across 45+ providers
- 20+ native guardrails plus LLM-based and partner checks
- Documented MCP gateway and mature observability
- AI features kept close to the product core
Cons
- Conditional routing is rule/parameter-based, not semantic
- Semantic caching is an Enterprise-tier feature
- RBAC (Production tier), SSO/SCIM, and VPC deploy are paid; no documented ensemble
Kong AI Gateway vs Portkey: feature comparison
Both pair an open-source core with a commercial tier, then diverge on form factor (API-gateway platform with add-on AI vs AI-native gateway with hosted governance) and where each draws the free/paid line.
| Feature | Kong | Portkey |
|---|---|---|
| Core & runtime | AI plugins on Kong Gateway (Apache-2.0 core, Lua/OpenResty) | Open-source TypeScript gateway (Node.js) + a separate hosted control plane |
| Provider coverage | 15+ providers (OpenAI, Azure OpenAI, Bedrock, Anthropic, Gemini, Vertex, Cohere, Mistral, and more) | 1,600+ models across 45+ providers |
| Routing | AI Proxy (single provider/model) in OSS; multi-model load balancing, semantic routing & failover via AI Proxy Advanced (Enterprise) | Load balancing, fallbacks, retries, conditional routing (metadata/params/path), circuit breaker |
| Semantic routing | ✓ Enterprise (AI Proxy Advanced) | — Rule/parameter-based only |
| Ensemble / fusion | — Not documented | — Not documented |
| Caching | AI Semantic Cache (Enterprise) | Simple + semantic caching (semantic = Enterprise) |
| Guardrails | AI Prompt Guard (regex) in OSS; AI Semantic Prompt Guard, AI PII Sanitizer, and provider guardrails (Azure Content Safety, AWS Guardrails, GCP Model Armor, Lakera) are Enterprise | 20+ deterministic native + LLM-based + partner checks; tiered |
| Rate limiting | Request-based in OSS; token- and cost-based via AI Rate Limiting Advanced (Enterprise) | Rate limits via Model Catalog; granular budgets Enterprise |
| Observability | Kong analytics and logging plugins; deeper analytics via Konnect | Logging, tracing, 21+ metric dashboard, OpenTelemetry-compliant; retention tiered |
| MCP gateway | AI MCP Proxy (Enterprise) | ✓ Documented (auth + access control for remote MCP) |
| LLM-as-judge | AI LLM-as-Judge (Enterprise) | — Not documented |
| Self-host / VPC | Self-host Docker/K8s (KIC), hybrid (data plane in your network), or Konnect SaaS | OSS self-hosts; managed VPC/hybrid is Enterprise |
| Enterprise identity | SSO (OIDC/SAML) + RBAC via Enterprise/Konnect; SCIM for Konnect not documented | RBAC from Production tier; SSO/SAML/OIDC & SCIM Enterprise |
Pricing comparison
Both have a free open-source core and paywall advanced features — but along different lines.
Kong Gateway's core is free (Apache-2.0) and several AI plugins are free, but the advanced AI plugins — multi-model load balancing, semantic routing and caching, token/cost rate limits, semantic and PII guardrails, MCP proxy, and LLM-as-judge — plus SSO and RBAC require the Enterprise tier (contact sales). Kong Konnect offers a free trial, a Plus tier with published usage pricing (around $200/month per additional 1M requests), and an Enterprise tier priced by sales. Portkey's gateway is free (MIT); the hosted platform is tiered — a free Developer tier, a paid Production tier (from about $49/month) that unlocks RBAC, and an Enterprise tier for SSO/SCIM, VPC/hybrid deployment, semantic caching, and granular budgets. In short, Kong gates advanced AI and identity to Enterprise, while Portkey keeps more AI features in lower tiers and gates enterprise identity, VPC, and semantic caching.
When to use Kong AI Gateway vs Portkey
Choose Kong to run AI plugins on a general-purpose API gateway you already operate; choose Portkey for an AI-native gateway with a hosted governance and observability UI.
Choose Kong AI Gateway if you…
- Already run (or want) a general-purpose API gateway like Kong
- Want AI plugins alongside your existing API traffic
- Are prepared to buy Enterprise for advanced AI and identity
Choose Portkey if you…
- Want the widest provider catalog behind one API
- Want a hosted observability, analytics, and governance UI
- Want more AI features (guardrails, MCP) closer to the product core
Bottom line
Choose Kong AI Gateway to add AI to a general-purpose API gateway (with advanced AI on Enterprise); choose Portkey for an AI-native gateway with hosted governance and 1,600+ models.
If you want AI plugins on a mature, general-purpose API-gateway platform — and can adopt Enterprise for multi-model routing, semantic features, and identity — Kong AI Gateway is the stronger pick; if you want an AI-native gateway with a hosted observability-and-governance UI and the widest provider catalog, Portkey fits better. If you're evaluating open-source AI gateways more broadly, AISIX is another option worth a look: a Rust, Apache-2.0 gateway from the creators of Apache APISIX, with semantic routing and ensemble built into the open-source core. See all AI gateway comparisons.
Frequently asked questions
Related comparisons
Portkey vs LiteLLM · AISIX vs LiteLLM · All AI gateway comparisons
Ready to get started?
For more information about full API lifecycle management, please contact us to Meet with our API Experts.
