A wrapper that lets you use the reverse-engineered Python library poe-api as if it was the OpenAI API for ChatGPT. You can connect your favorite OpenAI API based apps to this proxy and enjoy the ChatGPT API for free!
Poe.com from Quora is a free web app that lets you chat with GPT models. poe-api is a Python library that reverse-engineered poe.com so you can use Python to call poe. This project is a wrapper around poe-api that makes it accessible through an HTTP API, which mimics the official OpenAI API for ChatGPT so it can work with other programs that use OpenAI API for their features.
- Clone this repository to your local machine:
git clone https://github.com/juzeon/poe-openai-proxy.git
cd poe-openai-proxy/- Install dependencies from requirements.txt:
pip install -r external/requirements.txt- Create the configuration file in the root folder of the project according to the instructions in the comments:
vim config.tomlconfig.toml:
# The port number for the proxy service. The proxied OpenAI API endpoint will be: http://localhost:3700/v1/chat/completions
port = 3700
# A list of poe tokens. You can get them from the cookies on poe.com, they look like this: p-b=fdasac5a1dfa6%3D%3D
tokens = ["fdasac5a1dfa6%3D%3D","d84ef53ad5f132sa%3D%3D"]
# The proxy that will be used to connect to poe.com. Leave it blank if you do not use a proxy
proxy = "socks5h://127.0.0.1:7890"
# The gateway url for the Python backend of poe-api.
# Note that if you use docker this value should be changed into: http://external:5000
gateway = "http://127.0.0.1:5000"
# The gateway port for the Python backend of poe-api.
# Must be the same as the port indicated in the `gateway` above.
gateway-port = 5000
# The bot name to use from poe. `capybara` stands for `Sage`
bot = "capybara"
# Use leading prompts to indicate roles if enabled. You'd better disable it if you are using tools like https://github.com/TheR1D/shell_gpt
# 0:disable, 1:enable, 2:auto detect
# Example:
# ||>User:
# Hello!
# ||Assistant:
# Hello! How can I assist you today?
simulate-roles = 2
# Rate limit. Default to 10 api calls per token in 1 minute
rate-limit = 10
# Cool down of seconds. One same token cannot be used more than once in n seconds
cool-down = 3
# Timeout of seconds per response message
timeout = 200- Start the Python backend for
poe-api:
python external/api.py # Running on port 5000- Build and start the Go backend:
go build
chmod +x poe-openai-proxy
./poe-openai-proxyIf you would like to use docker, just run docker-compose up -d after creating config.toml according to the instructions above.
See OpenAI Document for more details on how to use the ChatGPT API.
Just replace https://api.openai.com/v1/chat/completions in your code with http://localhost:3700/v1/chat/completions and you're good to go.
Supported parameters:
| Parameter | Note |
|---|---|
| model | It doesn't matter what you pass here, poe will always use gpt-3.5-turbo. |
| messages | You can use this as in the official API, except for name. |
| stream | You can use this as in the official API. |
Other parameters will be ignored.