ChatGPT API Free Reverse MOBILE Proxy is a free reverse proxy to OpenAI API that allows users to access OpenAI API with an extra security layer.

• Features

• How to use ChatGPT API Reverse Proxy

  • Self-Host Your Own API
  • Use Our Hosted API
    • Text Completion
    • Chat Completion (ChatGPT)
    • Image Generation (DALL-E)
  • Examples using OpenAI libraries
    • Python
    • Node.js

• License

• Multiple OpenAI Keys - You can use multiple OpenAI keys. The API will randomly choose one of the keys to use.
• Moderation - The API has a built-in moderation system that will automatically check the prompt before sending it to OpenAI API (To prevent OpenAI terminate the account for violating OpenAI's policy).
• Streaming Response - The API supports streaming response, so you can get the response as soon as it's available.
• Same as Official - The API has the same endpoints as the official API, so you can use the same code to access the API (even the official OpenAI libraries)
• Free - The API is free to use through our hosted API (You can also self-host the API if you want).
• Extra mobile security layers - Extra middlewear checks (ex. UserAgent validation, MongoDB userId as request auth)

Note: Self-hosting it isn't free, you need to use your OpenAI Account credit.

In the past 6 months I have been trying securing my iOS App against any type of attack or OpenAI token leak, without much results that's why I decided to use a proxy and in specific to fork this one. So what I did and why?

What's special

As it's mobile proxy oriented, we don't care much about the White listed IP's (WHITELISTED_IPS) because mobile apps don't have a unique IP from where the requests comes from but instead we do have a very specific User Agent which usually contains the name of our app, I added an extra verification layer in which you can specify which is/are the allowed useer agents (ALLOWED_USER_AGENTS). That's the first extra security step mobile oriented.

Next! I do usually use MongoDB for store things related to the users, in specific in my app, when a new user is registered I do create a unique id which is persisting over deletion / re-install (you could use Firebase AUTH or you can just use some fingerprinting methods). Storing then this Unique User ID into MongoDB, you have a list of Users which use your app.

Based on this, we can send as mandatory "header" of the request the "userid" of the device which is sending the request, so we can keep track of who is sending the requests and how many requests are basically sent from the same "userid". This means that if you will notice TOO many requests from the same userid, most likely they just managed to understand how does the whole authorization works and you can just "remove" this userId from your MongoDB for stop allowing requests through your proxy. Easy! It's a sort of "token" auth, but in a more easy way.

As this is a special fork which uses MongoDB for iterate between a list of users and userid stored on your database / collection, you will definitely have to implement a way for save users from your app to your MongoDB and after that, you can just access from the proxy the collection and check if the user is present in the collection, if he is then we authorize the request to OpenAi, otherwise we decline it.

You can use ChatGPT API Reverse Proxy by choosing one of the following methods:

• Self-Host Your Own API
• Use Our Hosted API

‌

To self-host your own ChatGPT API, you can use the following steps:

1. Create an OpenAI API Key
2. Clone this repository and install the dependencies:

git clone https://github.com/PawanOsman/ChatGPT.git
cd ChatGPT
npm install

3. Set your OpenAI key and other configurations in the config.js file.
4. Start the server:

npm start

4. Use the API by sending an HTTP request to the API endpoints for example:

http://localhost:3000/v1/completions
http://localhost:3000/v1/chat/completions

To use our hosted ChatGPT API, you can use the following steps:

1. Join our Discord server.
2. Get your API key from the #Bot channel by sending /key command.
3. Use the API Key in your requests to the following endpoints.

https://api.pawan.krd/v1/completions

curl --location 'https://api.pawan.krd/v1/completions' \
--header 'Authorization: Bearer pk-***[OUR_API_KEY]***' \
--header 'Content-Type: application/json' \
--data '{
    "model": "text-davinci-003",
    "prompt": "Human: Hello\\nAI:",
    "temperature": 0.7,
    "max_tokens": 256,
    "stop": [
        "Human:",
        "AI:"
    ]
}'

https://api.pawan.krd/v1/chat/completions

curl --location 'https://api.pawan.krd/v1/chat/completions' \
--header 'Authorization: Bearer pk-***[OUR_API_KEY]***' \
--header 'Content-Type: application/json' \
--data '{
    "model": "gpt-3.5-turbo",
    "max_tokens": 100,
    "messages": [
        {
            "role": "system",
            "content": "You are an helpful assistant."
        },
        {
            "role": "user",
            "content": "Who are you?"
        }
    ]
}'

https://api.pawan.krd/v1/images/generations

curl --location 'https://api.pawan.krd/v1/images/generations' \
--header 'Authorization: Bearer pk-***[OUR_API_KEY]***' \
--header 'Content-Type: application/json' \
--data '{
    "prompt": "a photo of a happy corgi puppy sitting and facing forward, studio light, longshot.",
    "n": 1,
    "size": "1024x1024"
}'

You can use the same code to access the API using the official OpenAI libraries, the only difference is that you need to change the API key and the API base URL.

Examples are for text completion, but you can use the same code for chat completion and image generation.

You need to add the following lines before your code to use the API:

import openai

openai.api_key = 'pk-**********************************************'
openai.api_base = 'https://api.pawan.krd/v1'

Example code:

import openai

openai.api_key = 'pk-**********************************************'
openai.api_base = 'https://api.pawan.krd/v1'

response = openai.Completion.create(
  model="text-davinci-003",
  prompt="Human: Hello\nAI:",
  temperature=0.7,
  max_tokens=256,
  top_p=1,
  frequency_penalty=0,
  presence_penalty=0,
  stop=["Human: ", "AI: "]
)

print(response.choices[0].text)

You need to add the following lines before your code to use the API:

import { Configuration, OpenAIApi } from "openai";

const configuration = new Configuration({
	apiKey: "pk-**********************************************",
	basePath: "https://api.pawan.krd/v1",
});

Example code:

import { Configuration, OpenAIApi } from "openai";

const configuration = new Configuration({
	apiKey: "pk-**********************************************",
	basePath: "https://api.pawan.krd/v1",
});

const openai = new OpenAIApi(configuration);

const response = await openai.createCompletion({
	model: "text-davinci-003",
	prompt: "Human: Hello\nAI:",
	temperature: 0.7,
	max_tokens: 256,
	top_p: 1,
	frequency_penalty: 0,
	presence_penalty: 0,
	stop: ["Human: ", "AI: "],
});

console.log(response.data.choices[0].text);

This project is licensed under the MIT License - see the LICENSE file for details.