Tool

LibreChat

Local LLMs · Multi-provider chat UI · librechat.ai

An open-source, ChatGPT-shaped web UI you self-host — talk to many cloud providers and local endpoints (Ollama and friends) from one interface, with conversation history and team-oriented features.

What it is

LibreChat is a self-hosted web application for chatting with LLMs. It is not a native Windows app or AppImage; official docs steer Windows users toward Docker Desktop, and Linux installs via Docker, npm, or remote hosting guides. You wire API keys and/or local backends, then use a familiar chat UI with multi-conversation history. Source: github.com/danny-avila/LibreChat.

What it replaces

It replaces “everyone only uses the locked ChatGPT web UI” when you want one self-hosted front end for multi-model / multi-provider work — cloud APIs plus local runners — without handing the chat shell to a single vendor.

It does not replace a model runtime. LibreChat is the UI and orchestration layer; inference still comes from OpenAI-compatible APIs, Anthropic, Google, Azure, Ollama, vLLM, LiteLLM, and similar backends you configure. MIT code ≠ free hosted model tokens.

License

LibreChat is released under the MIT License, verified from the project’s primary LICENSE on GitHub (copyright year in that file may read 2026). Model weights, third-party APIs, and any hosted LibreChat offerings remain under their own terms and pricing.

Where it falls over

  • You still need API keys and/or a local backend — the UI alone generates nothing.
  • Docker/compose (or npm) ops: env files, secrets, upgrades, reverse proxy, and backups are on you.
  • Auth and secret hygiene matter once more than one person can hit the instance; don’t hang an open chat UI on the public internet with your keys.
  • Feature surface grows fast — upgrades can change config knobs; read release notes before jumping.
  • MIT covers the software, not model quality, rate limits, or cloud bills.
  • Not a substitute for Open WebUI / Jan / LM Studio when you want a thinner local-only path; pick based on multi-provider + self-host needs vs desktop simplicity.

Who it's for

Individuals and small teams who want a ChatGPT-shaped web UI they control, with multiple providers and optional local models behind it. Not a one-click “local AI” product if you skip wiring Ollama (or another runtime) and secrets.

Related in this cluster

  • Ollama — common local backend
  • Open WebUI — another self-hosted chat UI path (different license story)
  • Jan — desktop ChatGPT-shaped app
  • LiteLLM — gateway when you want one OpenAI-shaped proxy in front of many backends
  • vLLM — high-throughput GPU serving behind a UI
  • Local LLMs hub — runners, UIs, and hardware notes

Verdict

Use LibreChat when you want a MIT-licensed, self-hosted ChatGPT-shaped UI for multi-provider and local endpoints. Budget Docker/ops and API keys (or Ollama/vLLM/LiteLLM), harden auth before sharing, and remember the UI is not the model.

← Tools · Local LLMs · Home · librechat.ai · docs · GitHub