ChatGPT UI

English | 中文

A ChatGPT web client that supports multiple users, multiple database connections for persistent data storage, supports i18n. Provides Docker images and quick deployment scripts.

demo.mp4

📢Updates

2023-03-27

🚀 Support gpt-4 model. You can select the model in the "Model Parameters" of the front-end. The GPT-4 model requires whitelist access from OpenAI.

2023-03-23

Added web search capability to generate more relevant and up-to-date answers from ChatGPT! This feature is off by default, you can turn it on in `Chat->Settings` in the admin panel, there is a record `open_web_search` in Settings, set its value to True.

2023-03-15

Add "open_registration" setting option in the admin panel to control whether user registration is enabled. You can log in to the admin panel and find this setting option under Chat->Setting. The default value of this setting is True (allow user registration). If you do not need it, please change it to False.

2023-03-04

Update to the latest official chat model gpt-3.5-turbo

🎉🎉🎉Provide a shell script that can be used to quickly deploy the service to server Quick start

2023-02-24

Version 2 is a major update that separates the backend functionality as an independent project, hosted at [chatgpt-ui-server](https://github.com/WongSaang/chatgpt-ui-server).

If you still wish to use the old version, please visit the v1 branch.

Version 2 introduces the following new features:

• 😉 Separation of the frontend and backend, with the backend now using the Python-based Django framework.
• 😘 User authentication, supporting multiple users.
• 😀 Ability to store data in an external database (defaulting to Sqlite).
• 😎 Session persistence, allowing the API to answer questions based on your context.

🚀 One-click deployment

Note: This script has only been tested on Ubuntu Server 22.04 LTS.

bash <(curl -Ls https://raw.githubusercontent.com/WongSaang/chatgpt-ui/main/deployment.sh)

If you have a domain name, you can point it to the server's IP address using DNS resolution. Of course, using the server's IP address directly is also possible. During the script's execution, you will be prompted to enter a domain name. If you do not have a domain name, you can enter the server's IP address directly.

After the deployment is complete

Access http(s)://your.domain:9000/admin / IP http(s)://123.123.123.123:9000/admin to log in to the administration panel.

Default superuser: admin

Default password: password

Before you can start chatting, you need to add an OpenAI API key. In the Settings model, add a record with the name openai_api_key and the value as your API key.

Now you can access the web client at http(s)://your.domain or http://123.123.123.123 to start chatting.

🎉🎉🎉 Enjoy it!

Quick start with Docker Compose

Run services

Below is a docker-compose.yml template:

version: '3'
services:
  client:
    image: wongsaang/chatgpt-ui-client:latest
    environment:
      - SERVER_DOMAIN=http://backend-web-server
    depends_on:
      - backend-web-server
    ports:
      - '80:80'
    networks:
      - chatgpt_ui_network
  backend-wsgi-server:
    image: wongsaang/chatgpt-ui-wsgi-server:latest
    environment:
      - APP_DOMAIN=${APP_DOMAIN:-localhost:9000} # CSRF whitelist，Add the address of your chatgpt-ui-web-server here, default is localhost:9000
      - SERVER_WORKERS=3 # Number of gunicorn workers, default is 3
      #- DB_URL=postgres://postgres:postgrespw@localhost:49153/chatgpt # If this parameter is not set, the built-in Sqlite will be used by default. It should be noted that if you do not connect to an external database, the data will be lost after the container is destroyed.
      #- OPENAI_API_PROXY=https://openai.proxy.com/v1 # Proxy for https://api.openai.com/v1
      - DJANGO_SUPERUSER_USERNAME=admin # default superuser name
      - DJANGO_SUPERUSER_PASSWORD=password # default superuser password
      - [email protected] # default superuser email
      - ACCOUNT_EMAIL_VERIFICATION=none # Determines the e-mail verification method during signup – choose one of "none", "optional", or "mandatory". Default is "optional". If you don't need to verify the email, you can set it to "none".
      # If you want to use the email verification function, you need to configure the following parameters
    #      - EMAIL_HOST=SMTP server address
    #      - EMAIL_PORT=SMTP server port
    #      - EMAIL_HOST_USER=
    #      - EMAIL_HOST_PASSWORD=
    #      - EMAIL_USE_TLS=True
    #      - [email protected]  #Default sender email address
    ports:
      - '8000:8000'
    networks:
      - chatgpt_ui_network
  backend-web-server:
    image: wongsaang/chatgpt-ui-web-server:latest
    environment:
      - BACKEND_URL=http://backend-wsgi-server:8000
    ports:
      - '9000:80'
    depends_on:
      - backend-wsgi-server
    networks:
      - chatgpt_ui_network

networks:
  chatgpt_ui_network:
    driver: bridge

DB_URL schema

Engine	URL
PostgreSQL	postgres://USER:PASSWORD@HOST:PORT/NAME
MySQL	mysql://USER:PASSWORD@HOST:PORT/NAME
SQLite	sqlite:///PATH

Set API key

Access http(s)://your.domain:9000/admin / IP http(s)://123.123.123.123:9000/admin to log in to the administration panel.

Default superuser: admin

Default password: password

Before you can start chatting, you need to add an OpenAI API key. In the Settings model, add a record with the name openai_api_key and the value as your API key.

Now you can access the web client at http(s)://your.domain or http://123.123.123.123 to start chatting.

Donation

If it is helpful to you, it is also helping me.

If you want to support me, Buy me a coffee ❤️ https://www.buymeacoffee.com/WongSaang

Development

Setup

Make sure to install the dependencies:

# yarn
yarn install

Development Server

Start the development server on http://localhost:3000

yarn dev

Production

Build the application for production:

yarn build