ollama-html-ui

A HTML UI for Ollama.

Goals

• Minimal & responsive UI: mobile & desktop
• Secure: zero dependencies. Built using HTML, CSS, and JavaScript
• Cross browser: support last 2 versions with more than > 0.5% global use
• Simple installation: run in your browser, host on your own server. Less than 20 Kb gzipped

Screenshots

Desktop

Dark mode

Mobile

How to install Ollama

Manually

Download Ollama from the website.

MacOS

brew install ollama

Linux

For instructions on how to install Ollama on Linux, see https://ollama.ai/download/linux.

How to install and run the Ollama UI

To run the Ollama UI, all you need is a web server that serves dist/index.html and the bundled JS and CSS file.

First, start Ollama:

$ ollama run dolphin-phi

Using Caddy

To run the Ollama UI using Caddy, execute the following command from the command line:

$ git clone git@github.com:christianhellsten/ollama-html-ui.git
$ brew install caddy
$ caddy run

Using Docker

To run the Ollama UI using Docker, execute the following command from the command line:

$ docker run -p 80:80 aktagon/ollama-html-ui

Alternatively, build the image yourself:

$ git clone git@github.com:christianhellsten/ollama-html-ui.git
$ docker build -t ollama-html-ui .
$ docker run -p 80:80 ollama-html-ui

Using Parcel

First, clone the repository and install the dependencies:

$ git clone git@github.com:christianhellsten/ollama-html-ui.git
$ cd ollama-html-ui
$ yarn add --dev parcel
# Alternatively, use npm:
# npm install --save-dev parcel

Run the ollama-html-ui using parcel:

$ yarn parcel index.html

Open the UI using a browser:

$ open http://locahost:1234

Testing

Tests are written using Playwright and node:test.

The the tests can be run from the command line using this command:

$ ollama run dolphin-phi
$ node test

Deployment

$ parcel build index.html

Tasks

• Personas / Characters / Custom GPTs
• Ollama authentication
• Edit message / response
• Clear chat
• CSP
• Speech recognition
• Image upload / multi-modal
• Markdown support
• Fork chat
• Fork chat before / after message

Done

• Mark message as good, bad, flagged

• Export chat messages to JSON

• Keyboard shortcuts

• Dark & light theme

• Export chat messages to JSON

• Keyboard shortcuts

• Keyboard shortcuts

• Dark & light theme

• Delete message / response

• Ollama chat API / chat memory

• IndexedDB persistence

• Model parameters

• System prompt

• Copy message to clipboard

• Select model in settings (global)

• Select model in chat (local)

• Search chats

• Delete Chat

• Select model

• Save settings

• View settings

• Clear chats

• Edit chat

• New chat

• Abort response

• Send message

• UI tests: https://nodejs.org/api/test.html

Features

- https://ollama.ai support

**Chat**

- New chat
- Edit chat
- Delete chat
- Download chat
- Scroll to top/bottom
- Copy to clipboard

**Chat message**

- Delete chat message
- Copy to clipboard
- Mark as good, bad, or flagged

**Chats**

- Search chats
- Clear chats
- Chat history
- Export chats

**Settings**

- URL
- Model
- System prompt
- Model parameters

Troubleshooting

If you experience compilation errors, try deleting the cache directory:

```bash
rm -rf .parcel-cache/
```

Licensing

This project is available under two licensing options:

1. Open Source License (MIT):

   • The code in this project is available under the terms of the MIT License.
   • You are free to use, modify, and distribute the code in your non-commercial, open source projects.
   • View the full text of the MIT License in the LICENSE file.

2. Commercial License:

   • If you intend to use this code in a commercial project, we offer a separate commercial licensing option.
   • Our commercial license provides additional rights and support tailored to your commercial needs.
   • To inquire about our commercial licensing options, pricing, and terms, please contact us at christian@aktagon.com to discuss your specific requirements.

We value and support both our open source community and commercial users. By providing dual licensing options, we aim to make this project accessible to a wide range of users while offering customized solutions for commercial projects.

Built with

HTML, CSS, and JavaScript