LiteLLM

LiteLLM란 무엇인가요?

LiteLLM is an LLM gateway, also described as an OpenAI proxy, built to manage authentication, load balancing, and spend tracking across more than 100 LLMs. The product positions itself as a way to give developers model access through the OpenAI format while keeping control in a central proxy.

The site frames LiteLLM as useful for platform teams and organizations that need to manage access for many developers and projects. It supports model fallbacks, spend tracking, budgets, rate limits, guardrails, and observability features, with both open source and enterprise offerings available.

LiteLLM(으)로 무엇을 할 수 있나요?

OpenAI-compatible gateway

Provide a single OpenAI-compatible gateway for calling models across more than 100 LLMs, while keeping the same request format across providers.

Spend tracking and cost attribution

Track usage by key, user, team, or organization, and attribute spend across providers such as OpenAI, Azure, Bedrock, and GCP.

LLM fallbacks and load balancing

Route requests across providers to handle outages and reuse multiple deployments under the same model name when needed.

Budgets and rate limits

Apply budgets and rate limits, including RPM and TPM tracking, to help manage usage across teams and deployments.

Operational controls

Add guardrails, prompt management, and pass-through endpoints to shape how requests move through the proxy.

Logging and observability

Use logging and observability features such as s3 logging, OTEL logging, and integrations like Langfuse, Arize Phoenix, and LangSmith.

사용 사례

“Standardize model access”

Centralize access to multiple model providers behind a single proxy so developers can call models without learning provider-specific formats.

“Handle outages and rate limits”

Route traffic to alternate providers or deployments when a model is unavailable or rate-limited, helping teams keep applications running.

“Manage spend across teams”

Track usage by team, org, key, or user so platform owners can attribute cost and monitor consumption across projects.

“Control access in larger organizations”

Use enterprise identity and admin controls when many developers need temporary or group-based access to models.

“Self-managed enterprise deployment”

Deploy the gateway in a self-managed environment when the organization needs its own infrastructure or upgrade assistance.

자주 묻는 질문

What is LiteLLM?

LiteLLM is an LLM gateway / OpenAI proxy that helps teams manage authentication, load balancing, and spend tracking across 100+ LLMs. The source describes it as being available in the OpenAI format.

Can LiteLLM be self-hosted?

The source indicates that LiteLLM can be deployed on-prem and that LiteLLM Enterprise is available as a secure, self-managed deployment.

Does LiteLLM have both open source and enterprise options?

Yes. The home page describes Open Source as free and the enterprise page describes Enterprise as a separate offering with features such as SSO, SCIM, OIDC/JWT auth, support, and live upgrade assistance.

Who is LiteLLM intended for?

The published source does not provide a full setup guide, but it emphasizes OpenAI-compatible access, model access control, and enterprise identity features for organizations that need to manage access for many developers and projects.

주요 정보