Another ChatGPT Web

Introduction

This project is developed from the source project chatgpt-web with backend refactored from third-party chatgpt-api which has been out-of-maintenance to official openai-node API with much flexibility and extensibility.

New Features

This project has the following new features standing on the source project chatgpt-web.

Frontend

[✓] Model selection (gpt-4 / gpt-3.5)

[✓] In-place message editing

[✓] Full session history control

[✓] Conversation-wise draft prompt and context state

[✓] Search chat list by title

[✓] Drag to reorder chat list

[✓] Auto-generate summary title for conversation

Backend

[✓] Built with official openai-node API

[✓] Multi-user authentication and profiles

[✓] Model-wise token usage statistics logging for users

More coming...

Prerequisites

Node

node >= 18

PNPM

If you haven't installed pnpm

npm install -g pnpm

Filling in service/.env file

You should complete the environment variable file following service/.env.example, including setting up your OpenAI API Key.

[optional] Prepare service/auth.json file

If you would like to setup access tokens for users, create the auth config file following service/auth.example.json to define relevant user information, including name, avatar link, and description.

Install Dependencies

Under root directory of the repository:

Frontend

pnpm install

Backend

cd service/

pnpm install

Run in Test Environment

Start backend service

In one terminal:

cd service/

pnpm dev

Start frontend proxy

In another terminal:

pnpm dev

Access http://127.0.0.1:3002/ to see a preview during development.

Packaging and Deployment

Frontend Webpage

Optionally modify frontend .env file.

# Install
pnpm install

# Pack - this packs all distribution files into the backend folder "service/public/"
pnpm build

Backend Service

After packing the frontend, you can safely move service/ folder independently to the production environment.

Setup service/.env and service/auth.json files as described above.

cd service/

# Install
pnpm install

# Pack
pnpm build

# Start service
pnpm prod

Token Usage Logging

User usage logs would be generated automatically under service/logs/, for example:

{
  "2024_06_28": {
    "user1": {
      "gpt-4o": {
        "prompt_tokens": 768,
        "completion_tokens": 266
      },
      "gpt-4o-mini": {
        "prompt_tokens": 324,
        "completion_tokens": 95
      }
    }
  }
}

FAQ

Q: Does it support unofficial proxy API like in source project?

A: No. It removes proxy support with official API.

Q: Can I migrate my chat history from the source project?

A: Yes. The data store is fully compatible with the source project, and you can safely import your old chat histories.

Q: Does it support other packaging methods like docker container?

A: To maintain simplicity, this project only keeps manual deployment. You could apply other packaging methods from the source repo.

Contributing

Thanks to everyone who has contributed! Especially @Gnefil who learnt a lot during the project development =D.

Acknowledgements

Thanks to @ChenZhaoYu for contributing to the source project chatgpt-web.

License

MIT © [ChenZhaoYu, RyanXinOne]