LibreChat
LibreChatSelf-hosted chat interface with native support for OpenAI, Anthropic, Google, Mistral and local model backends, plus multi-user authentication, agents, a code interpreter and prompt sharing. It aims at a faithful reproduction of the ChatGPT interaction model with enterprise authentication and role-based access, which suits org-wide deployments. ClickHouse acquired the project in 2025.
Verified 2026-08-13 via the danny-avila/LibreChat repository and the ClickHouse acquisition announcement.
Openness
5 high confidence- license
- MIT(OSI)
- source
- public
- core-gated
- ungated
Fully OSI (MIT), full source public, no paid feature-gated core. Enterprise backing via ClickHouse acquisition has not introduced a closed core. The code interpreter it ships points at the open ClickHouse/code-interpreter project rather than a paid tier. The acquisition post itself names no license; MIT comes from the repository.
- https://github.com/danny-avila/LibreChat recorded 2026-08-13
MIT license; full-featured self-hosted chat UI
- https://clickhouse.com/blog/clickhouse-acquires-librechat recorded 2026-08-13
ClickHouse acquires LibreChat (Nov 4, 2025); LibreChat remains open source/MIT, no closed core introduced, ongoing OSS investment
Adoption
3 medium confidenceThe quantity banded here is GitHub stars on danny-avila/LibreChat: 42,002 read from the GitHub API, which is >10K stars and level 3 on the stars scale. A stars-based band cannot exceed level 3 however large the count, because a star is attention rather than use. Stars are the only countable signal this product has. Docker Hub accounts for 372,272 pulls on the primary librechat/librechat repository and roughly 732,090 across the whole librechat organisation; the image people actually run is on GHCR, which publishes no pull count at all; and the vendor's own counter claims 47.3M pulls, two orders of magnitude above what any registry can produce and checkable against nothing. The vendor's star counter, by contrast, matches the GitHub API almost exactly (42k against 41,995), so the counter is live rather than invented - it is the pull number specifically that has no reachable basis. Deployment reach is certainly higher than 42k; nothing published measures it.
- https://api.github.com/repos/danny-avila/LibreChat recorded 2026-08-14
GitHub API for danny-avila/LibreChat - 41,995 stars, 8,686 forks, 201 watchers, MIT. A countable signal; no user or deployment count is published anywhere.
- https://api.github.com/repos/danny-avila/LibreChat recorded 2026-08-14
GitHub API for danny-avila/LibreChat - 42,002 stargazers, 8,687 forks, 201 watchers, MIT. This is the figure the level 3 band is taken on. The digest differs from the other read of the same endpoint because the star and fork counts moved.
- https://www.librechat.ai/ recorded 2026-08-14
vendor counter reading "42k GitHub Stars", "47.3M Docker Pulls", "387 Contributors". The star figure matches the GitHub API; the pull figure is roughly 65x what Docker Hub's own API accounts for across the whole librechat org and cannot be checked, since the primary image is on GHCR.
- https://github.com/danny-avila/LibreChat recorded 2026-06-04
MIT self-hosted chat platform with broad provider support. Carries no Docker-pull figure, so it never supported the 23M this band was placed on.
Capability
5 high confidenceFrontier feature surface for the category: code interpreter, MCP, agents/subagents and artifacts on top of the standard chat-UI matrix. At/above Open WebUI on breadth.
- https://github.com/danny-avila/LibreChat recorded 2026-08-13
feature list: multi-provider, code interpreter, MCP, agents/subagents, artifacts, OAuth2/LDAP
Verified 2026-08-13