Your cookie choices

We use necessary cookies to make this site work. With your permission, analytics cookies help us understand usage and marketing cookies support advertising.

You can reject non-essential cookies and still use the site. Change your choices anytime in Cookie settings.

Cookie settingsCookie policy

Kong AI Gateway review

Updated September 28, 2026 · DeployCue Editorial Team · Facts checked September 26, 2026

We earn a commission when you sign up through links on this site. Partner status can affect where a brand appears in our rankings. Partners with a live offer in your country come first, then every other brand by score.

Our verdict

Kong AI Gateway suits platform teams that already run Kong and want a governed proxy layer in front of many providers. Bring your own key and an OpenAI-compatible API keep migration light, but a per-gateway subscription plus a monthly charge per unique LLM proxied and no free tier put it ahead of LiteLLM for projects.

Key facts

Best for
Teams already on Kong who need one policy layer across 19 LLM providers.
Fee model
Per-gateway subscription plus USD 100/month per unique LLM proxied↗
Providers routed
19↗
Bring your own key
Yes↗
OpenAI-compatible API
Yes↗
Free tier
No↗
Legal entity
Kong Inc.↗
Headquarters
United States↗
Support channels
Email, Chat, Help-centre, Community↗
Official website
konghq.com

Facts checked September 26, 2026, 3 sources

Apps and profiles

Elsewhere
LinkedInX

Pros and cons

Pros

  • Bring your own key is supported, so model spend stays with the provider rather than the gateway.
  • The API is OpenAI-compatible, so existing OpenAI client code can point at the gateway as it stands.
  • It routes 19 providers, which reaches the 19 upper range of this group.
  • Support runs on email, chat and help centre, so most questions are answerable without a sales call.
  • Kong Inc., headquartered in the US, is a named legal entity rather than an anonymous service.

Cons

  • Billing is a per-gateway subscription plus a monthly charge per unique LLM proxied, and LiteLLM's fee model is open-source self-host that stays free under an MIT licence with an enterprise annual licence priced to annual gateway request capacity and never per token.
  • There is no free tier, so evaluation runs on a paid gateway, while LiteLLM's fee model is open-source self-host that stays free under an MIT licence.
  • Support runs on email, chat and help centre, so a phone line is not among the listed channels.
  • At 19 providers routed the coverage is narrower than the largest gateways in this group, which reach 140.
Kong AI Gateway

Ready to try Kong AI Gateway?

Scored 8.7 / 10 by DeployCue.

See alternatives

Who it suits

  • Platform and API teams that already run Kong and want LLM traffic governed by the same policies.
  • Developers who hold their own provider keys and want an OpenAI-compatible endpoint over many providers.
  • Organisations that prefer a subscription with a named legal entity in the US over a community project.

Who should look elsewhere

  • Solo developers testing a single provider, who can self-host LiteLLM for free under an MIT licence.
  • Buyers who need the broadest provider coverage in this group, where UCloud lists as few as from 4 while the group reaches 140.
  • Teams that want a flat, predictable subscription, since Salad bills Flat monthly subscription or pay per token.

What it connects to

Kong AI Gateway exposes an OpenAI-compatible API, so OpenAI client code can be pointed at the gateway without rewriting requests. Bring your own key is supported, which means the model provider bills the call directly and the gateway acts as the policy layer rather than the buyer of tokens. That split suits teams that already hold provider contracts and want routing, logging and access control gathered in a single place. The practical limit is provider coverage, given in the features section below.

Pricing

The fee model is a per-gateway subscription plus a monthly charge for every unique LLM proxied through it. The charge follows distinct providers, not tokens or requests, so a team routing many models under a single plan pays more than a team routing a single upstream. There is no free tier, so evaluation starts on a paid gateway rather than on a trial allowance. For contrast, LiteLLM's fee model is open-source self-host that stays free under an MIT licence, with an enterprise annual licence priced to annual gateway request capacity and never per token, and Salad lists Flat monthly subscription or pay per token.

What it does well

The gateway routes 19 providers, which sits at the upper end of what most gateways in this group cover, and that range matters for teams that want a failover path rather than a single upstream. Bring your own key keeps provider contracts with the model vendors, and the OpenAI-compatible API means a single endpoint can sit in front of that range of providers. The trade-off is that the OpenAI-compatible surface is a common denominator, so provider specific features behind an upstream API may need extra gateway configuration.

Who runs it

The service is operated by Kong Inc., registered in the US under the terms of use. Support runs on email, chat and help centre. Self-serve routes suit developers reading documentation, while email and chat cover questions about the per-gateway subscription or the per unique LLM charge. A named company also gives procurement teams a real counterparty on the subscription.

How Kong AI Gateway compares

Frequently asked questions

What does Kong AI Gateway cost?+

The fee model is a per-gateway subscription plus a monthly charge for each unique LLM proxied through it. The bill follows the number of distinct providers you route rather than the volume of calls, so each extra provider adds a charge on top of the plan. LiteLLM's fee model is open-source self-host that stays free under an MIT licence.

Is there a free tier or free trial for Kong AI Gateway?+

There is no free tier, so any evaluation of the gateway runs on a paid plan from the first gateway onward. Budget for the per-gateway subscription plus the monthly per unique LLM charge before starting a proof of concept, and note that 19 providers are routable behind that plan. LiteLLM's self-host path stays free under an MIT licence if you only need to test routing.

Can I use my own provider API keys with Kong AI Gateway?+

Bring your own key is supported, so your model providers bill you directly and the provider contracts stay in your own name. The gateway adds the routing and policy layer on top rather than reselling tokens across the 19 providers it routes. That split suits teams already holding provider accounts and explains the appeal against gateways that bill per token.

How many LLM providers can Kong AI Gateway route?+

It routes 19 providers, which is at the upper end of the range most gateways in this group cover. That headroom gives room for a failover path when an upstream model is unavailable. Coverage still stops well short of the largest gateways in the group, which reach 140.

Do I need to change my code to use Kong AI Gateway?+

The API is OpenAI-compatible, so existing OpenAI client code can be pointed at the gateway endpoint instead of a provider endpoint, and that holds across the 19 providers it routes. The trade-off is that a shared compatible surface is a common denominator, so provider specific features behind an upstream API may need extra gateway configuration.

Who runs Kong AI Gateway and where is it based?+

The service is operated by Kong Inc., a named company registered in the US under its terms of use. A named counterparty matters for procurement review and for teams that need an invoice from a real company rather than a community project. The per-gateway subscription sits with that same entity.

What support channels does Kong AI Gateway offer?+

Support runs on email, chat and help centre, so email and chat sit alongside a help centre and community answers. Self-serve routes cover most documentation questions, while email and chat suit questions about the subscription or the per unique LLM charge. A phone line is not among the listed channels.

Is Kong AI Gateway right for a small team or a side project?+

A per-gateway subscription plus a monthly charge per unique LLM proxied is heavier for a small team than a self-hosted option, and with no free tier there is no free way to try it. LiteLLM fits that case better under an MIT licence. Kong AI Gateway earns its cost once several of the 19 providers and shared policy control are already in play.

Sources

  • developer.konghq.com/ai-gateway/ai-providers checked September 26, 2026

    Backs: Bring your own key, OpenAI-compatible API, Providers routed

    “Centralized AI Provider credential management”

  • konghq.com/legal/terms-of-use checked September 26, 2026

    Backs: Headquarters, Legal entity

    “Kong administers and operates the Website from its headquarters in San Francisco, California USA.”

  • konghq.com/pricing checked September 26, 2026

    Backs: Fee model, Free tier, Support channels

    “Unique LLMs proxied by AI plugins prorated hourly. No costs for 30 days $100/month per model”

Compiled from 3 source pages, checked September 26, 2026. We did not open an account.

Kong AI Gateway

Kong AI Gateway

Teams already on Kong who need one policy layer across 19 LLM providers.

8.7/10See alternatives