Call 100+ LLMs in the standard OpenAI format with open-source load balancing and budget guardrails.
Open-source edition has zero feature caps for self-hosting on local servers or Kubernetes.
LiteLLM is the universal proxy and SDK connecting applications to 100+ LLMs using the standard OpenAI format. Whether routing to Anthropic, AWS Bedrock, Google Vertex AI, Azure OpenAI, or self-hosted Ollama instances, LiteLLM translates requests seamlessly. It ships with enterprise-grade features out of the box: generate virtual API keys, set spend budgets with alerts, enforce RPM/TPM rate limits, and load balance traffic across redundant deployments.