Taken from LiteLLM's own pricing page.
| Plan | Price | Included |
|---|---|---|
| Open source | Free forever, self-hosted | 140+ provider integrations, virtual keys, budgets and teams, load balancing, RPM and TPM limits, LLM guardrails, logging into Langfuse, Arize Phoenix, LangSmith and OTEL. MIT licensed, no credit card. |
| Enterprise | Request pricing | Everything above plus JWT auth, SSO, audit logs, enterprise support and custom SLAs. |
This is where most comparisons cheat in one direction or the other. Vendors inflate the cost of self-hosting to sell you a subscription. Open source advocates pretend it is zero. Neither is honest.
What you actually take on: a service in your infrastructure that sits in the path of every AI call your product makes. You deploy it, you patch it, you monitor it, and when it falls over at an awkward hour, someone on your team owns that. If you already run services and have somewhere to put this, that is a small marginal cost. If LiteLLM would be the only thing you self-host, it is not.
I am not going to put an hours figure on that, because it depends entirely on infrastructure you already have and I have not measured yours. Work it out against your own team before you decide.
✓ You already run your own services and adding one more is routine
✓ You want no vendor between your application and the model providers
✓ Data residency or compliance means the traffic cannot leave your infrastructure
✓ You want to read and change the routing logic yourself
✓ You have no platform team and no appetite to become one
✓ You want log retention and dashboards without building them
✓ A monthly fee is genuinely cheaper than the attention self-hosting takes. Portkey and Helicone both have free tiers to test that against before you spend anything
Every tool on this page is built for engineers. They answer which model handled a request, how long it took, and what it cost in aggregate. That is the right shape for a product team.
It is the wrong shape for an agency. If you are billing eleven clients and running AI inside deliverables for all of them, the month-end question is not what your OpenAI bill was. It is which client is eating your margin. That means tying every call to a client, a deliverable and a rate card, then putting it somewhere your billing process can read. A gateway gives you raw usage. It does not give you attribution, and it does not put it on an invoice.
That is the part I build. Not a gateway, and not a product with a monthly fee. A cost attribution and reporting layer that sits on top of whichever gateway you already picked, wired into how you actually bill.
If your AI bill is fine but your client margins are not, that is a different problem. Worth 30 minutes.
Talk through your AI costs →