Verified 2026-11-18 · rates follow the live catalog
LiteLLM is an MIT-licensed gateway and Python SDK that you host: one OpenAI-format API over 100+ providers, with virtual keys, budgets, spend tracking and fallbacks. You bring your own provider keys and run your own Postgres and Redis, so there is no token markup — and nobody else to call when it breaks. An Enterprise license adds SSO and SCIM, audit logs and support, priced by request capacity rather than per token. Openference is the opposite trade: hosted, with usage bundled into plans, so there is nothing to operate.
Self-hosted software · Free (MIT) · Enterprise by quote · Checked 19 November 2026 against LiteLLM Pricing.
Pick LiteLLM when the gateway has to run in your own cloud, or in front of your own provider accounts. Pick Openference when you would rather not operate a gateway and want one monthly price.
| Plan | Price | What it includes |
|---|---|---|
| Open Source | $0 (self-hosted) | 100+ providers behind one OpenAI-format API, virtual keys, users and teams, spend tracking, budgets and rate limits, fallbacks, logging and Prometheus metrics. |
| Enterprise | Annual license, quoted | Everything in open source plus SSO and SCIM, OIDC/JWT auth, audit logs, secret managers and key rotation, org and team admins, multi-region control and 24/7 support. Sized by request capacity, never per token. SSO is free for up to five users. |
| LiteLLM | Openference | |
|---|---|---|
| Where it runs | Your infrastructure — Docker, Kubernetes or air-gapped | Hosted by Openference |
| Pricing model | Free open source; Enterprise is an annual license sized by request capacity | Plan-based (included usage) + on-demand credits |
| Model access | You bring your own provider keys (100+ providers) | Models included behind one Openference key |
| Fee on top of model rates | None — you pay each provider directly | No per-request fee and no markup on listed rates; 5% processing fee on credit top-ups |
| Automatic failover across providers | Yes — fallbacks you configure in the router | Yes |
| Operations | You run the database, cache and upgrades | Nothing to operate |
| Open-weight models | Yes | Yes |
| Proprietary frontier models | Yes | Yes |
Free (MIT) · Enterprise by quote. The plan table above lists each tier as LiteLLM publishes it, read from LiteLLM Pricing on 19 November 2026.
That depends on how you use it. Pick LiteLLM when the gateway has to run in your own cloud, or in front of your own provider accounts. Pick Openference when you would rather not operate a gateway and want one monthly price.
LiteLLM's from its own pages on 19 November 2026; Openference's from its live catalog and public plans. Both change, so confirm on the linked page before you commit.
More on the comparison index, the current plans and the model catalog.
Model catalog · Pricing · Comparisons · Cost guides · Integration guides · FAQ
Guides: GLM coding plan alternative · Free AI API · LLM gateway · LLM proxy · Cheap AI API · OpenAI-compatible API
Model APIs: GLM API · DeepSeek API · Kimi API · Qwen API · MiniMax API