Chat UI
Chat UI is Hugging Face's open-source chat interface under Apache-2.0, the SvelteKit app that powers HuggingChat. It talks to any service that speaks the OpenAI protocol, including a llama.cpp server, Ollama through its OpenAI-compatible bridge, and OpenRouter, stores chats in MongoDB, calls tools on MCP servers, and has an optional router that picks a model for each message. It runs with npm or Docker.
What does Chat UI do?
Chat UI is "a chat interface for LLMs. It is a SvelteKit app and it powers the HuggingChat app on hf.co/chat." It supports only OpenAI-compatible APIs, set through one base URL and the models endpoint, and the README says "any service that speaks the OpenAI protocol (llama.cpp server, Ollama, OpenRouter, etc.) will work by default." Hugging Face's Inference Providers router and Poe are other examples.
Chats, users, settings, files and stats live in MongoDB, with an embedded MongoDB saving to a local folder if you do not set one. It calls tools exposed by MCP servers, which administrators can preset and users can add; web search is one such server. An optional router called Omni picks a model per message: a multimodal model for image input, an agentic one for tool use, and a default for the rest. Image input is supported per model. The README says provider-specific integrations, embeddings and built-in web-search helpers were removed, and the old version sits on a legacy branch.
Key facts
- Licence: Apache-2.0, copyright The Hugging Face team, with no added conditions.
- Powers HuggingChat at hf.co/chat.
- Supports only OpenAI-compatible APIs through OPENAI_BASE_URL and the models endpoint; llama.cpp server, Ollama's OpenAI-compatible bridge, OpenRouter, Poe and Hugging Face's router are named.
- Chats, users, settings, files and stats in MongoDB 6 or 7, with an embedded fallback that saves to ./db.
- MCP tools from preset or user-added servers; web search through an MCP server such as Exa.
- Omni router: multimodal for image inputs, agentic for tool use, default otherwise; you supply the routing file.
- Run: npm install and npm run dev, or the Docker image chat-ui-db that bundles MongoDB.
- Latest tagged release when read: v0.10.0 on 2026-05-11; 10,953 stars on 2026-09-17.
What does it replace, and where does it fall short?
Chat UI replaces ChatGPT's interface with the one Hugging Face runs in production, pointed at any OpenAI-compatible model you choose, local or hosted, on your own server. It suits someone who wants a clean, maintained front end with MCP tools and little else.
Where it falls short: it speaks only the OpenAI protocol, so every model needs an OpenAI-compatible endpoint. Document chat and embeddings were removed, the README does not describe sign-in setup, image generation or voice, and the router ships with no sample configuration. For a feature-complete platform, Open WebUI or LibreChat on the open-source alternatives to ChatGPT page.
How does Chat UI run?
Clone, npm install and npm run dev, which serves on port 5173, then npm run build for production; deploying may need an adapter for your target. Or run the Docker image that bundles MongoDB. Set OPENAI_BASE_URL and a key to an OpenAI-compatible service. Hugging Face runs the same code as HuggingChat; the README describes no paid tier.
Who is Chat UI for?
A developer who wants HuggingChat's interface on their own server, in front of a llama.cpp or Ollama endpoint, with MCP tools. A team that needs document chat, sign-in and roles should pick Open WebUI or LibreChat.
What limits does the README state?
From the README: "Chat UI only supports OpenAI-compatible APIs"; provider-specific integrations, GGUF discovery, embeddings and web-search helpers "are removed"; no sample router file ships; deploying may need an adapter.
Questions people ask
Is Chat UI open source?
Yes. It is Apache-2.0 with no added conditions, copyright the Hugging Face team. It is the same codebase that runs HuggingChat at hf.co/chat, and the README keeps an older version on a legacy branch.
Can Chat UI use local models?
Yes, through any server that speaks the OpenAI protocol. The README names a llama.cpp server and Ollama with its OpenAI-compatible bridge as examples. Chat UI itself does not run models; it connects to whatever endpoint you give it.
Does Chat UI support document chat?
No. The README says embeddings and related provider-specific features were removed from the current version. Web search and other tools come through MCP servers instead. For document chat, AnythingLLM or Open WebUI are the options on this site.
How does Chat UI compare with ChatGPT?
It gives you HuggingChat's interface on your own server, with any OpenAI-compatible model and MCP tools, for free. ChatGPT includes OpenAI's models, documents, images and voice as a service. The open-source alternatives to ChatGPT page compares it with the fuller platforms.
What is the Omni router in Chat UI?
An optional server-side router that picks a model for each message: a multimodal model when the message includes an image, an agentic model when tools are needed, and a default model otherwise. You define the routes in a JSON file yourself, since the README says no sample ships.
Sources
- Chat UI README and Apache-2.0 LICENSE: github.com/huggingface/chat-ui, read 2026-09-17.
Compared with the others
On the open-source alternatives to ChatGPT page, Chat UI is the pick for huggingchat's interface on your own server. Apache-2.0 SvelteKit app that powers HuggingChat, for any OpenAI-compatible endpoint, with MongoDB storage, MCP tools and a per-message model router.
Also on that page: Open WebUI for a private chatgpt for a whole team, LibreChat for every provider in one mit platform, AnythingLLM for chatting with your documents.
More chat interfaces
A self-hosted chat platform for Ollama and OpenAI-compatible models with document search over 9 vector databases, web search, image generation, roles and single sign-on, under a BSD-3 licence with an added branding clause.
LobeHub, formerly LobeChat, is an agent workspace with an agent builder, agent groups, scheduled runs, editable memory and a library of over 10,000 tools and MCP plugins, under a community licence that needs a paid licence to distribute a modified version.
An MIT all-in-one AI app from Mintplex Labs for chatting with your documents with citations, running agents with MCP, and using any of 39 model providers including local llama.cpp and Ollama, as a desktop app or a multi-user Docker server.
An AGPL-3.0 desktop client for Windows, macOS and Linux that connects OpenAI, Gemini, Anthropic and local Ollama or LM Studio models, with 300+ preset assistants, side-by-side model chats, file support and MCP servers.
An AGPL-3.0 desktop app for local models, formerly text-generation-webui, with five engines including llama.cpp and ExLlamaV3, an OpenAI and Anthropic-compatible API, tool calling, file chat, image generation and LoRA training, fully offline.
An Apache-2.0 desktop app from Menlo Research that downloads and runs open models from Hugging Face on your own computer, connects to OpenAI, Anthropic and others when you want, serves a local OpenAI-compatible API, and supports MCP.
Added September 17, 2026. Every claim above comes from the project's README, LICENSE or model card, read on September 17, 2026, or from the GitHub API on the date shown in the panel. Found an error? Write to reveneau@licheo.com and it is fixed in the next weekly pass. Repository: github.com/huggingface/chat-ui.
Running one of these inside your own environment, with your own data and your own security rules, is the kind of work Reveneau does. Read how a forward deployed engagement works.