Open-source AI infrastructure
Six alternatives to LiteLLM compared on licence terms, which capabilities need a paid key when self-hosting, and published pricing. Facts checked against each proje…
They have hit the enterprise line. LiteLLM's root LICENSE places everything under enterprise/ under a separate licence, and virtual keys, budgets, RPM and TPM limits, spend by key, user, team and org, SSO with SCIM and OIDC auth sit there. Enterprise pricing is not published, so it arrives as a sales conversation rather than a line item you can plan for.
Bifrost gates guardrails, cluster mode, adaptive load balancing, RBAC and audit logs. Kong gates seventeen AI plugins including semantic caching and every guardrail. Helicone gates supported on-prem deployment. Everstack gates none of these. Find the line before adopting, because governance and high availability are what you need in month six.
LiteLLM has the widest provider coverage of anything self-hostable, describing itself as calling 100+ LLM APIs. One niche provider you depend on can block a full cutover, and finding that out in week three is expensive. Also run a grep for `import litellm`: calling the proxy over HTTP moves with a base URL change, but using the Python library as a library is a rewrite.