chatgpt-to-api's Introduction

ChatGPT-to-API

Create a fake API using ChatGPT's website

IMPORTANT

You will not get free support for this repository. This was made for my own personal use and documentation will continue to be limited as I don't really need documentation. You will find more detailed documentation in the Chinese docs by a contributor.

API endpoint: http://127.0.0.1:8080/v1/chat/completions.

中文文档（Chinese Docs）

Setup

Authentication

Access token retrieval has been automated by OpenAIAuth with account email & password.

accounts.txt - A list of accounts separated by new line

Format:

email:password
...

All authenticated access tokens will store in access_tokens.json

Auto renew access tokens after 14 days

Caution! please use unblocked ip for authentication, first login to https://chat.openai.com/ to check ip availability if you can.

API Authentication (Optional)

Custom API keys for this fake API, just like OpenAI api

api_keys.txt - A list of API keys separated by new line

Format:

sk-123456
88888888
...

Getting set up

git clone https://github.com/xqdoo00o/ChatGPT-to-API
cd ChatGPT-to-API
go build
./freechatgpt

Environment variables

PUID - A cookie found on chat.openai.com for Plus users. This gets around Cloudflare rate limits
SERVER_HOST - Set to 127.0.0.1 by default
SERVER_PORT - Set to 8080 by default
OPENAI_EMAIL and OPENAI_PASSWORD - It will automatically refresh your PUID if set (requires Plus account)
ENABLE_HISTORY - Set to true by default

Files (Optional)

proxies.txt - A list of proxies separated by new line
```
http://127.0.0.1:8888
...
```
access_tokens.json - A JSON array of access tokens for cycling (Alternatively, send a PATCH request to the correct endpoint)
```
["access_token1", "access_token2"...]
```