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

LiteLLM 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

LiteLLM suits teams that want one gateway across many model providers without paying a gateway tax. The self-host build is free forever under an MIT licence, and the Enterprise plan is priced on annual gateway request capacity rather than tokens.

Key facts

Best for
Self-hosting a multi-provider LLM gateway on your own infrastructure.
Fee model
Open-source self-host free forever ($0, MIT licence); Enterprise annual licence priced to annual gateway request capacity, never per token↗
Providers routed
140↗
Bring your own key
Yes↗
OpenAI-compatible API
Yes↗
Free tier
Yes↗
Legal entity
Berrie AI Incorporated↗
Headquarters
United States↗
Founded
2023↗
Support channels
Email, Help-centre, Community↗
Official website
litellm.ai

Facts checked September 26, 2026, 4 sources

Apps and profiles

Elsewhere
LinkedInXgithub.comdiscord.comyoutube.com

Pros and cons

Pros

  • Self-hosting is free forever under an MIT licence, and a free tier is available.
  • Bring your own key, so you hold the provider contracts and set the terms.
  • An OpenAI-compatible API means existing client code needs no rewrite.
  • The Enterprise licence is priced to annual gateway request capacity and never per token, so heavy routing does not raise token bills.
  • 140 providers routed in one endpoint, well above the 9 median across the category.

Cons

  • Support runs on email, help centre and community only, so a paid response has no phone or live chat route.
  • The Enterprise annual licence is priced to request capacity, which means a quote is needed before you know the cost.
  • A self-hosted free forever build puts the hosting, patching and uptime work on your own team.
  • The fee model covers two routes: a free forever self-host build under an MIT licence, and an Enterprise licence quoted on annual gateway request capacity.
LiteLLM

Ready to try LiteLLM?

Scored 9.1 / 10 by DeployCue.

See alternatives

Who it suits

  • Engineers running several model providers behind one internal endpoint.
  • Teams that must keep provider keys in their own environment rather than with a gateway vendor.
  • Organisations standardising on an OpenAI-compatible client interface.

Who should look elsewhere

  • Teams wanting a managed gateway with a support line: LiteLLM support runs on email, help centre and community.
  • Buyers who need a fixed posted price: the Enterprise licence is priced to annual gateway request capacity.

Integrations and platforms

LiteLLM exposes an OpenAI-compatible API, so clients, SDKs and internal tools written against that interface can point at the gateway without code changes. That covers the largest integration win most teams look for: one endpoint shape that keeps working as providers come and go. Bring your own key is supported, which means the provider accounts, contracts and rate limits stay in your organisation rather than with a gateway vendor. The trade-off is that you own the runtime: LiteLLM is software you deploy, so the routing layer inherits whatever uptime discipline your infrastructure has. There is no hosted platform layer to fall back on in the free build, which is the reason some teams pair it with their own deployment tooling.

Core features

The core feature is coverage: 140 providers routed through a single interface, which is the widest spread in this comparison and a long way above the 9 median for gateways. Bring your own key means the gateway is a routing and observability layer rather than a resale point, so spend goes to the model providers you already use. The OpenAI-compatible API keeps the call format fixed while the underlying provider can change, which is the practical way to run fallbacks across vendors. Pricing follows the same split. The self-host build is free forever under an MIT licence and a free tier exists, while the Enterprise annual licence is priced to annual gateway request capacity and never per token, so heavier routing does not quietly inflate model spend.

The company behind it

LiteLLM is run by Berrie AI Incorporated, founded in 2023 and headquartered in the United States. The self-host build is released under an MIT licence, so the code can be inspected, forked and hosted by the buying team, and the legal entity behind it is named on the vendor record rather than left anonymous. Support for the project runs on email, help centre and community. That mix suits teams who want to own the deployment and read the source, and it is the trade-off for anyone who needs a phone line or live chat for an outage. A young company with an open licence means the software moves quickly, and the commercial backing for it is the annual Enterprise licence.

How LiteLLM compares

Frequently asked questions

Is LiteLLM free to run?+

Yes. The open-source self-host build is free forever under an MIT licence, and a free tier is also listed. Paid spend comes only if you choose the Enterprise annual licence, which is priced to annual gateway request capacity and never per token.

What does LiteLLM charge for the Enterprise version?+

The Enterprise plan is an annual licence priced to your annual gateway request capacity, never per token. Because the price is quoted against capacity rather than listed per seat, you would need a quote before setting a budget for it.

How many AI providers can LiteLLM route to?+

LiteLLM routes 140 providers, the widest coverage in this comparison and far above the 9 median for gateway providers. That breadth lets one endpoint hold several vendors and fall back between them.

Can I use my own provider API keys?+

Yes, bring your own key is supported. Provider accounts, contracts and rate limits stay with your organisation, and the gateway acts as a routing and observability layer rather than a resale point for model usage.

Do I have to change my code to use LiteLLM?+

Usually not. LiteLLM offers an OpenAI-compatible API, so clients, SDKs and internal tools built against that interface can be repointed at the gateway without rewriting the call format. That is what makes a provider swap cheap.

Who runs LiteLLM and where are they based?+

LiteLLM is operated by Berrie AI Incorporated, founded in 2023 and headquartered in the United States. The named legal entity is published on the vendor record, which matters for procurement and security review.

What support can I get from LiteLLM?+

Support runs on email, help centre and community. There is a help centre for documentation, email for direct questions and a community for shared problems, so a live chat or phone escalation is not part of the mix.

Is LiteLLM a managed service or do I host it?+

You host it yourself in the free build, which means your team owns the deployment, patching and uptime of the routing layer. Teams that do not want that work are usually better served by a hosted gateway with a support line behind it.

Sources

  • docs.litellm.ai/docs/data_security checked September 26, 2026

    Backs: Legal entity, Support channels

    “Legal Entity Name: Berrie AI Incorporated”

  • litellm.ai checked September 26, 2026

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

    “Bring your own internal, fine-tuned, and self-hosted models behind the same key.”

  • litellm.ai/pricing checked September 26, 2026

    Backs: Fee model, Free tier

    “The open-source gateway is free to self-host, forever. Add Enterprise when you need governance, security, and support - priced to your gateway's scale, never per token.”

  • linkedin.com/company/litellm checked September 26, 2026

    Backs: Founded, Headquarters

    “Founded 2023”

Compiled from 4 source pages, checked September 26, 2026. We did not open an account.

LiteLLM

LiteLLM

Self-hosting a multi-provider LLM gateway on your own infrastructure.

9.1/10See alternatives