Guides · Updated
Running a local AI chat with a nice interface (Open WebUI and alternatives)
Get a ChatGPT-style interface for your local models: setting up Open WebUI with Ollama, plus simpler desktop apps and browser extensions if you'd rather skip Docker.
Chatting in a terminal works, but most people want chat history, file uploads and a model dropdown. You can get all of that on top of local models. Everything below runs on your own machine and connects to Ollama or runs models itself.
Quick answer
- Want the full ChatGPT-style experience in a browser, possibly for several people? Use Open WebUI.
- Want a single desktop app with no setup? Use LM Studio or Jan, or the chat window in the Ollama app.
- Mostly chatting with documents? AnythingLLM is built around that.
Open WebUI with Ollama
Open WebUI is a self-hosted web interface. You open it in your browser, but it runs on your PC. It supports multiple chats, model switching, document upload, voice input and user accounts.
Step 1: Install Ollama and a model
Follow the guide for Windows, Mac or Linux, then pull a model, for example:
ollama pull gemma4:e4b
Not sure what fits your GPU? Use the GPU picker.
Step 2: Install Open WebUI
Option A: Docker (recommended). Install Docker Desktop (Windows/Mac) or Docker Engine (Linux), then run:
docker run -d -p 127.0.0.1:3000:8080 --add-host=host.docker.internal:host-gateway -v open-webui:/app/backend/data --name open-webui --restart always ghcr.io/open-webui/open-webui:main
127.0.0.1:3000means only your own PC can open it. The official example uses-p 3000:8080, which also lets other devices on your network reach it.--add-hostlets the container reach Ollama running on your PC.-v open-webui:...keeps your chats and settings in a Docker volume, so they survive updates.
Option B: Python. If you’d rather not use Docker, install it with pip (the project recommends Python 3.11):
pip install open-webui
open-webui serve
This version runs on port 8080 instead of 3000.
Step 3: Create the admin account
Open http://localhost:3000 (or :8080 for the pip install). The first account you create becomes the administrator. The account is stored locally; it’s how Open WebUI separates users, not a cloud sign-up.
Your Ollama models should appear in the model dropdown. If the list is empty, go to Admin Panel → Settings → Connections and check the Ollama URL: http://host.docker.internal:11434 from Docker, or http://localhost:11434 for the pip install. Still stuck? See When it doesn’t work.
Step 4: Useful features to try
- Documents: attach a file to a chat, or create a Knowledge collection to reuse. See Chat with your documents locally.
- Voice: dictate with the microphone button. Speech options are covered in Local speech-to-text and text-to-speech. For a spoken back-and-forth that stays on your PC, see Build a private voice assistant on your own PC.
- Images: connect Open WebUI to a local ComfyUI install in Admin Panel → Settings → Images to generate pictures from your chats. See Generating images locally.
- System prompts and custom models: save a model with your own instructions, such as “answer in plain English” or “reply as a code reviewer”.
Updating
With Docker, pull the new image, then remove and recreate the container with the same docker run command. Your data stays in the open-webui volume:
docker pull ghcr.io/open-webui/open-webui:main
docker rm -f open-webui
With pip: pip install -U open-webui.
Alternatives
| App | Type | Good for |
|---|---|---|
| Ollama app | Desktop (Windows, Mac) | Simplest option: a chat window built into Ollama itself |
| LM Studio | Desktop | Browsing and downloading models, document chat, no Docker |
| Jan | Desktop, open source | A clean offline chat app that can run models itself or connect to Ollama |
| AnythingLLM | Desktop or self-hosted | Document workspaces with an Ollama connector |
| LibreChat | Self-hosted web app | Teams mixing local and cloud models in one interface |
| Page Assist | Browser extension | A sidebar chat for Ollama inside Chrome or Firefox |
Comparing the two most popular options in more depth: LM Studio vs Ollama. For independent scores across more tools, see VibeEngine’s local LLM rankings.
Keep it private
- Most of these apps can also connect to cloud providers. Chats with cloud models leave your machine; chats with local models don’t.
- Features like web search send your questions to outside services. Leave them off if that matters to you.
- Don’t expose Open WebUI or Ollama to the internet. To use it from your phone or laptop, use a VPN such as Tailscale.
The full checklist is in Local AI for privacy, step by step. The documented off switch for web search, call mode, cloud connections and the version check is in What leaves your PC?.
Sources: Open WebUI docs, Ollama docs.
Get the Local AI weekly
New models worth running, what fits which GPU, and setup fixes. One short email a week, free. Unsubscribe any time.
Local-LLM App Starter
A tested starter for building your own app on Ollama: chat, RAG and a GPU-fit checker. Skip the setup hours.
kits.lilaforge.com → VibeEngineBest local LLM tools, ranked
Ollama, LM Studio, Open WebUI and llama.cpp scored from public evidence, with top picks.
vibeengine.lilaforge.com →