Skip to content
Atlas by Comet Lab
Surface layer

AI chat interfaces, compared

A chat interface is the window people type into. It sits on top of a model so nobody has to build a front end, and the fuller ones add accounts, history and chat over your documents. The first question is who it serves: a team, from your own servers, or one person, on their own machine.

Open this layer in the explorer

At a glance

Every tool in this layer, with where it fits and the facts that usually decide it.

Tool Where it fits Licence Self-hostable Hosted API Maturity
Open WebUI Self-hosted chat front-end The default choice when a team wants a private ChatGPT-style assistant inside its own perimeter, with accounts and document chat, and no front-end to build. Open source Yes No Established
AnythingLLM Document chat app Best when non-technical users need to chat with their own documents quickly, and you want retrieval handled for you rather than assembled. Open source Yes No Growing
big-AGI Self-hosted chat front-end A fit for technical users who want to experiment across models and compare answers in one interface. Open source Yes No Growing
Chatbox Local desktop app Best for an individual who wants a single, simple desktop client across several model providers. Open source Yes No Established
Hollama Minimal chat front-end A fit when you want the lightest possible self-hosted chat window over a model endpoint, with nothing to operate. Open source Yes No Emerging
Jan Local desktop app A fit for individuals or teams who want a local AI assistant with a hard guarantee that conversations never leave the machine. Open source Yes No Growing
LibreChat Self-hosted chat front-end A strong pick when you want a familiar chat UX across several providers and expect to customise the codebase rather than take it as-is. Open source Yes No Established
LobeChat Self-hosted chat front-end A good choice when presentation matters and you want a self-hosted assistant that feels like a finished consumer product. Open source Yes No Established
Msty Local desktop app A good pick when a non-technical user wants local models running with the least possible setup. Proprietary Yes No Growing
TypingMind Bring-your-own-key front-end A fit for individuals or small teams who already hold provider keys and want a better interface without a recurring per-seat fee. Proprietary No No Established

How to choose

Start from your situation. Where a line says “our”, it comes from Comet Lab’s own builds.

  1. A private assistant for a team, with no front end to build

    Accounts and document chat out of the box, inside your own perimeter. Our usual starting point for an internal assistant.

  2. You want to own and change the front-end code

    A familiar chat UX across several providers, built to be customised rather than taken as it is. We pick it over Open WebUI when the client wants to shape the code.

  3. Presentation matters

    A self-hosted assistant that feels like a finished consumer product, with plugins and custom agents.

  4. People need to chat with their own documents, quickly

    Retrieval is handled for you. If the pilot succeeds and retrieval quality starts to matter, expect to move to a pipeline you control.

  5. One person, on their own machine

    Jan keeps conversations on the machine, Msty has the least setup, and Chatbox puts several providers in one desktop client.

  6. Comparing models side by side

    Several models answer at once, for technical users who want to push models hard and compare them.