18870 / chatgpt-proxy

A ChatGPT unofficial API proxy written in Python.
GNU Affero General Public License v3.0
15 stars 4 forks source link

Notice

This project does NOT works until further notice

ChatGPT-Proxy

Python version of OpenAI's ChatGPT web API proxy
Python alternative of ChatGPT-Proxy-V4
Use cookie cf_clearance to pass Cloudflare browser check

_puid no longer works

Requirements

Install

pip install chatgpt-proxy

Usage

Run as a service

Set these environment variables:

Or create a .env file with your environment variables at where you want to run the proxy:

cf_clearance=YOUR_CF_CLEARANCE
user_agent=YOUR_USER_AGENT
access_token=YOUR_ACCESS_TOKEN
puid=ANY_VALID_PUID
proxy_trust_client=False
host=127.0.0.1
port=7800
mod_access_token=YOUR_MOD_ACCESS_TOKEN

Note that environment variables will override the values in .env file.

Then run: python -m chatgpt_proxy

Success

If you see this in console: 2023-01-01 00:00:00,000 - chatgpt_proxy.proxy - INFO - puid: user-xxxxxx You are ready to go

The proxy is avaliable at http://host:port/backend-api/

Integrate into your FastAPI app

Check out __main__.py

from chatgpt_proxy import WebChatGPTProxy
proxy = WebChatGPTProxy(cf_clearance=CF_CLEARANCE, user_agent=USER_AGENT, access_token=ACCESS_TOKEN, trust=False)
app = FastAPI()
proxy.attach(app, path="/backend-api")

class WebChatGPTProxy:

Behind a http proxy

You need to use the same ip address you used to get the cookie cf_clearance

Set HTTP_PROXY and HTTPS_PROXY or ALL_PROXY environment variables.
This cannot be set in .env file becauce httpx (the package we used to send request) reads from environment variables only. See also httpx#Proxies.

Credits

License

This work is licensed under the GNU Affero General Public License v3.0 or later, with the "CHATGPT-PROXY" exception.

"CHATGPT-PROXY" EXCEPTION TO THE AGPL

As a special exception, using this work in the following ways does not cause your program to be covered by the AGPL:

  1. Bundling the unaltered code or binary of this work in your program; or
  2. Interacting with this work through the provided inter-process communication interface, such as the HTTP API.