Local & self-hosted apps

Decision matrix

Chatbot UI app decision matrix
Overview

Chatbot UI is a free, MIT-licensed chat interface you self-host and point at OpenAI, Azure or a local Ollama model using your own API keys. The vendor's own hosted demo at chatbotui.com is currently offline, so self-hosting with Supabase and Vercel is the only working route. Pick it when you want a familiar chat window without a subscription and without a vendor picking the model for you.

Other good fits
  • A small team that wants to self-host a chat front end so conversations stay in infrastructure they control.
Pros
Read pros

Chatbot UI is an open-source (MIT-licensed) chat interface: instead of a vendor's own chat app, you get a self-hostable front end you point at a model provider using your own API key. It supports OpenAI and Azure endpoints, plus local models through Ollama. The project's own hosted demo at chatbotui.com is currently offline (it returns a dead-deployment error), so self-hosting is the only working way to use it today.

Pick it when…

  • You want a plain chat window without a monthly subscription, paying the model provider directly per token instead.
  • You want to switch between providers (OpenAI, Azure, a local Ollama model) from one interface instead of separate apps.
  • You're comfortable bringing your own API key and self-hosting, since the vendor's own hosted option isn't currently available.
Cons
Read cons

Bring-your-own-key means you're responsible for that key's security and spend limits; a leaked key on a self-hosted instance is your problem to fix, not a vendor's. It's a chat interface, not a document workspace or agent platform, so if you want citations, retrieval over your own files, or autonomous agent behaviour, you'll need a different tool alongside or instead of it.

Pick something else when…

  • You want retrieval over your own documents with citations. AnythingLLM or Open WebUI are built for that.
  • You don't want to manage API keys or hosting at all. A managed assistant like ChatGPT or Claude trades that setup for a subscription.
  • You want a workflow builder around model calls, not just a chat window. Dify adds that layer.

Avoid if

  • You want document workspaces, citations or agent features built in; Chatbot UI is a chat window, not a retrieval or agent platform.
Workflow

Self-host with Node.js, Docker and a Supabase backend for the database, deployed on something like Vercel → add your own API key for OpenAI, Azure or point it at a local Ollama model → chat as you would with any hosted assistant → switch models per conversation depending on the task and its cost.

Cost

Chatbot UI itself is free software; you pay only the model provider for tokens used, which is often cheaper than a flat monthly subscription for light use, but can cost more for heavy daily use where a subscription's fixed price wins. Self-hosting adds a Supabase account (for the backend database) and Vercel hosting (for the frontend) to the mix, both of which have their own free tiers and paid plans; check their current pricing separately since Chatbot UI's own licence doesn't cover them.

Restrictions

Self-hosting needs Node.js, Docker and a Supabase account for the database; the project's own hosted demo is currently offline, so there's no working no-setup option right now. MIT licensing means you can modify and redistribute it, but you take on the security and maintenance of whatever you deploy.

First project

Documentation-based suggestion — not run here.

Set up a self-hosted instance with your own OpenAI API key, then compare one everyday task, drafting a short email, on two different models available through your key. Note the difference in response quality against the per-token cost of each.

Done when you've picked a default model for routine use based on that comparison, not just on which one you'd heard of first.

Links & evidence
Documentation-based · not hands-on · Sep 23, 2026

Rests on Chatbot UI's GitHub repository, read on 2026-09-23. No instance was installed or run for this review. The vendor's own hosted demo at chatbotui.com returned a dead-deployment error (Vercel DEPLOYMENT_NOT_FOUND) when checked, so this review treats self-hosting as the only current route.

  • Chatbot UI GitHub repository · Sep 23, 2026 — MIT license, self-host requirements (Node.js, Docker, Supabase), OpenAI/Azure/Ollama support.
  • Open WebUI documentation · Sep 23, 2026 — Comparison alternative with a stronger local-model focus.
  • AnythingLLM documentation · Sep 23, 2026 — Comparison alternative for document workspaces and citations.
Further review notes & alternatives

Alternatives and trade-offs

  • Open WebUI wins when you want a similar bring-your-own-model chat front end with a stronger focus on local models.
  • AnythingLLM wins when document workspaces and citations matter more than a plain chat window.
  • ChatGPT wins when you'd rather pay a flat subscription and skip API key management entirely.

Learn more

Catalogue facts

Provide a simple web chat interface for language models you configure. Choose it when an open-source Next.js frontend fits your deployment plan. It can run locally or on Node.js hosting with a Supabase backend.

free · self-host

  • chat frontend
  • byok
  • next.js
  • supabase
  • self-hosted
  • talk-think
  • open-source
View sourced image

Chatbot UI

Press Escape to close.

Chatbot UI workspace screenshot from the official repository README. The empty chat state and model label reflect the documented version.

Chatbot UI — mckaywrigley/chatbot-ui README

Owner-approved use of public official interface screenshots, as recorded in public-visuals-batch-1.json, 13 September 2026. Copyright retained by McKay Wrigley and Chatbot UI contributors; attribution retained. Editorial product illustration; no open licence asserted. Converted on 15 October 2026 to a 1600px-wide WebP (quality 80) for faster page loads; the original image remains in the media library.

Original image source (opens in a new tab)

Browse all tools