Skip to content

LibreChat Review

LibreChat is a free, MIT-licensed AI chat platform you self-host. It puts OpenAI, Anthropic, Google, local models and 30+ other providers behind one interface with agents, MCP, SSO and search.

Free · API

Stork's verdict on LibreChat

LibreChat is the strongest free choice for teams that want one self-hosted chat for every model, with SSO and agents included, as long as someone can run the server.

LibreChat reviewed by Stork AI · stork.ai/en/librechat

Screenshots

stork
I am a painting.
One still painting and one typed sentence.
This painting was never filmed.
A photo, a cartoon, a painting or a 3D character. From 10 cents a minute. Free to try.
Make a photo talkDevelopers: there is an API

What LibreChat does

LibreChat is an open-source web app that gives a team one chat interface for many AI models. On its own site it lists OpenAI, Anthropic, Google, Azure OpenAI, AWS Bedrock, Mistral, xAI, DeepSeek, Groq, OpenRouter and Perplexity among its providers, plus self-hosted backends such as Ollama, vLLM and LiteLLM through custom endpoints. On top of chat it adds agents (with files, tools, reusable skills and subagents), MCP connections to outside tools, Artifacts that turn replies into interactive apps, diagrams and documents, persistent memory, projects, web search with citations, search across your message history, and a code interpreter. ClickHouse announced on November 4, 2025 that it had acquired LibreChat, and said the project stays 100% open source under its existing MIT license.

There is no desktop installer and no LibreChat-run SaaS plan: you deploy it yourself and use it in a browser. LibreChat's docs offer Docker Compose (recommended; it bundles MongoDB, MeiliSearch, the RAG API and a vector database), npm with Node.js, or a Helm chart for Kubernetes, and list DigitalOcean, Railway and Hugging Face among hosting options. The stated minimum is 1 GiB RAM and 1 vCPU, with 2 GB suggested when everything is enabled. Admins configure models and features in librechat.yaml or, since 2026, in a separate Admin Panel web app for users, groups, roles and configuration. A public demo runs at chat.librechat.ai.

LibreChat pricing

LibreChat itself costs nothing. Its about page says it is MIT licensed with "no subscriptions, no restrictions", and there is no paid tier or license key.

PlanMonthly priceIncluded
LibreChat (self-hosted, MIT)$0Full app: agents, MCP, Artifacts, code interpreter, web search, memory, SSO (OAuth, SAML, LDAP), 2FA, moderation, Admin Panel
ClickHouse Agents (managed, built on LibreChat)$0Public beta in ClickHouse Cloud "at no additional charge"; ClickHouse says pricing will be available soon

What you do pay for when self-hosting:

  • Model usage: each provider (OpenAI, Anthropic, Google and so on) bills its own API usage. Local models through Ollama or vLLM cost only your hardware.
  • Hosting: a server or cloud instance with at least 1 GiB RAM and 1 vCPU (2 GB suggested with everything enabled), plus MongoDB and the bundled search and RAG services.
  • Optional services: web search uses third-party providers you configure (Serper, SearXNG, Tavily or Keenable for search; Firecrawl and others for scraping; Jina or Cohere for reranking), each with its own pricing.

The hosted LibreChat Code Interpreter API at code.librechat.ai, once a paid add-on, shows "Subscriptions Currently Unavailable" and is not accepting new subscribers; ClickHouse said in 2025 it would discontinue the API in its current form, targeting May 1, 2026. LibreChat's docs now point to the open-source code-interpreter service (Apache 2.0) that you deploy next to LibreChat. ClickHouse Agents needs a ClickHouse Cloud account, which is billed separately by ClickHouse.

Prices checked on the vendor's pricing page on November 10, 2026. LibreChat has no pricing page; see About LibreChat for the license terms and ClickHouse Agents for the managed option.

Key features of LibreChat

What LibreChat's site and documentation describe:

  • One interface for 30+ providers and local models via custom OpenAI-compatible endpoints
  • Agents with files, tools, reusable skills and delegated subagents, shareable between users
  • MCP support to connect agents to outside tools and internal services
  • Code interpreter in a sandbox: Python, Node.js, Go, C/C++, Java, PHP, Rust and Fortran
  • Web search with citations through Serper, SearXNG, Tavily or Keenable, plus rerankers
  • Artifacts, persistent memory, projects, chat forking and message search
  • SSO via OAuth, SAML and LDAP, two-factor auth, rate limiting and moderation
  • Admin Panel web app for roles, groups, permissions and config without editing YAML

Who LibreChat is best for

LibreChat suits companies, universities and technical teams that want an internal ChatGPT-style workspace under their own control: one place where staff can use several commercial models and in-house agents, with SSO and role-based access, while chats stay on infrastructure they run. Its homepage lists organizations including Shopify, Daimler Truck, Stanford and NASA among users. You need someone comfortable running Docker or Kubernetes, MongoDB and API keys, and keeping it updated.

If you mainly run local models on one machine and want the quickest start, Open WebUI is the common self-hosted alternative, and Jan or AnythingLLM run as desktop apps with no server at all. If you just want a personal multi-model client with your own API keys, Chatbox is simpler. Teams already on ClickHouse Cloud who would rather not operate anything can try ClickHouse Agents, the managed service built on LibreChat.

What users say about LibreChat

Large deployments are the strongest public signal. Daimler Truck (September 2025) says it chose LibreChat because it is open source and adaptable, runs it on AWS so entered data stays in the company, and that several thousand pilot users built over 800 of their own AI solutions and agents. In ClickHouse's acquisition post, which is the owner's own announcement, a Shopify engineer describes near-universal internal adoption with thousands of custom agents. TrueFoundry's comparison with Open WebUI (updated October 2026; TrueFoundry sells a competing AI governance layer) credits LibreChat's MIT license, provider breadth and unpaywalled SAML and LDAP, but notes that local models need custom endpoint configuration, RAG means running a separate RAG API and vector database, and free self-hosting still costs time for databases, updates and uptime.

Pros

  • MIT licensed with no paid tier, so every feature is free to self-host
  • Connects 30+ cloud providers and local models in one interface
  • SSO with OAuth, SAML and LDAP plus 2FA included without a paywall
  • Agents, MCP, code interpreter, web search and memory built in
  • Used at scale by organizations such as Daimler Truck and Shopify

Cons

  • Self-hosted only; you run MongoDB, search and updates yourself
  • Local models and RAG need extra configuration and services
  • Hosted Code Interpreter API closed to new subscribers
  • Managed option (ClickHouse Agents) requires a ClickHouse Cloud account

LibreChat FAQ

Is LibreChat free?

Yes. LibreChat is open source under the MIT license with no paid tier or subscription. You pay only for your own hosting and for the AI providers whose APIs you connect; local models through Ollama or vLLM add no API cost.

Is there a hosted or cloud version of LibreChat?

LibreChat itself is self-hosted only; its docs say there is no desktop installer and you run it with Docker, npm or Helm. The closest managed option is ClickHouse Agents, which ClickHouse says is built on LibreChat and is in public beta for ClickHouse Cloud customers at no additional charge, with pricing to come. A public demo runs at chat.librechat.ai.

Who owns LibreChat?

ClickHouse announced on November 4, 2025 that it had acquired LibreChat. It said LibreChat remains 100% open source under its MIT license, that existing deployments keep working unchanged, and that it uses LibreChat internally for data analysis.

Does LibreChat work with Ollama and local models?

Yes. LibreChat lists Ollama, vLLM, LiteLLM and Apple MLX among supported backends, connected as custom OpenAI-compatible endpoints in librechat.yaml. You can mix them with cloud providers in the same interface.

LibreChat vs Open WebUI: which should I pick?

Both are self-hosted AI chat interfaces. LibreChat leans toward multi-provider teams, with SAML, LDAP and 2FA built in and an MIT license. Open WebUI is often chosen for local-model and RAG-centred setups. TrueFoundry's comparison (written by a vendor of a competing governance layer) frames it as a choice of operating priorities rather than a clear winner.

What happened to the LibreChat Code Interpreter API?

The hosted Code Interpreter API at code.librechat.ai was a paid service. Its page now says subscriptions are unavailable and it is not accepting new subscribers; ClickHouse said it would discontinue the API in its current form, targeting May 1, 2026. LibreChat's docs now use an open-source code-interpreter service (Apache 2.0) that admins deploy themselves.

What do I need to run LibreChat?

LibreChat's docs list a minimum of 1 GiB RAM and 1 vCPU (2 GB suggested with everything enabled). The recommended route is Docker Compose, which bundles MongoDB, MeiliSearch, the RAG API and a vector database; npm needs Node.js and your own MongoDB, and a Helm chart is available for Kubernetes.

Compare Alternatives

LibreChat

Free · API

From the Stork blog

Enjoying this? Get one like it in your inbox each morning.

one email a day · unsubscribe in two clicks · no third-party tracking