Best Open-Source AI Chatbots 2026: 6 Options Compared

An open-source AI chatbot is usually two things stacked together: a chat interface that you own and a model that answers. The interfaces listed here are open source, so you can run them on a laptop or a server and keep conversations off third-party sites. The model is a separate choice. You can point them at a hosted API such as OpenAI or Anthropic and pay per use, or at a local model through Ollama or llama.cpp and pay with your own hardware. LibreChat and Onyx are built for teams, with sign-in and shared access to many providers. Jan and GPT4All are desktop apps that a single user can install in minutes. AnythingLLM adds document chat on top of either route. Flowise is different: it is a builder for chat flows, closer to a toolkit than a finished assistant. The main trade-off is effort against control. A hosted chat service works on day one and is better at the hardest questions, while a self-hosted setup takes an afternoon and may need a decent GPU to feel fast. Read each project's license, since a few mix an open core with a separately licensed enterprise folder, and check whether the model you plan to load allows your use.

LibreChat

MIT-licensed chat interface that connects to many providers and is self-hosted. It is free, and you pay only the model providers you connect, which suits a team that wants one shared front end.

Price: Free (Free and self-hosted; you pay the model providers you connect) (LibreChat pricing, checked 2026-10-22).

Platforms and license: Self-hosted, Web. MIT (open source) (license source).

Project activity: 46.4k GitHub stars · last commit Oct 27, 2026

Best for: Teams sharing one self-hosted chat.

AnythingLLM

MIT desktop app and Docker self-host, both free, with document chat built in. A hosted Basic plan exists at $50 a month if you do not want to run it yourself.

Price: Free (Desktop app free; self-hosting with Docker free; no limits stated); paid from $50 per month, Basic hosted plan billed monthly (Pro is $99 per month; Enterprise custom) (AnythingLLM pricing, checked 2026-10-22).

Platforms and license: Desktop, Self-hosted via Docker. MIT (open source) (license source).

Project activity: 67.7k GitHub stars · latest release v1.19.0 (Oct 26, 2026) · last commit Oct 27, 2026

Best for: Chatting with your own documents.

Compare: AnythingLLM alternatives

Jan

Apache-2.0 desktop app that runs models on your machine. The home page describes it as free and open source, and no paid prices were shown.

Price: Free (Jan pricing, checked 2026-10-22).

Platforms and license: Desktop. Apache-2.0 (open source) (license source).

Project activity: 45.1k GitHub stars · latest release v0.8.7 (Oct 20, 2026) · last commit Oct 27, 2026

Best for: Single users who want a local app.

GPT4All

MIT-licensed desktop app from Nomic that runs models on your own machine with no account. Check the vendor page for supported platforms.

Price: Free (Free open-source desktop app that runs models on your own machine) (GPT4All pricing, checked 2026-10-22).

Platforms and license: Windows, macOS, Linux. MIT (open source) (license source).

Project activity: 77.4k GitHub stars · latest release v3.10.0 (Feb 25, 2025) · No commits since May 27, 2025

Best for: Private offline chat on a laptop.

Onyx

MIT Community Edition you can self-host, built around search across connected apps. The repository lists 50 or more connectors; a Lite mode cannot index documents.

Price: Free (Community Edition is free to self-host; Onyx Cloud offers a free trial) (Onyx pricing, checked 2026-10-23).

Platforms and license: Self-hosted (Docker, Kubernetes), Onyx Cloud. MIT (Community Edition); Enterprise Edition adds paid features (open source) (license source).

Project activity: 33k GitHub stars · latest release v5.0.1 (Oct 26, 2026) · last commit Oct 28, 2026

Best for: Company knowledge chat.

Flowise

Apache-2.0 visual builder for LLM flows with chat as one output. The core is free; enterprise directories sit under a separate commercial license, so read the license split.

Price: Free (Flowise pricing, checked 2026-10-22).

Platforms and license: Self-hosted (Node.js, Docker). Apache-2.0; enterprise code under a commercial license (open source) (license source).

Project activity: 55.5k GitHub stars · latest release flowise@3.1.4 (Jul 29, 2026) · Archived on GitHub

Best for: Building custom chat flows without much code.

Worth knowing: Node-based visual builder for LLM flows. Free under Apache-2.0 for the core; enterprise directories are under a separate commercial license.

Which one to pick

• You are one person and want a local app: Jan
• You run a team and want sign-in: LibreChat
• You want to chat with PDFs and notes: AnythingLLM
• You need company-wide search and chat: Onyx
• You want to design the flow yourself: Flowise