Skip to main content

A reverse engineered API wrapper for OpenPlayground (nat.dev)

Project description

Python OpenPlayground API

PyPi Version

This is an unoffical API wrapper for the website OpenPlayground, which provides access to a wide array of AI models for free, including ChatGPT, GPT-4, and Claude.

Notice:

OpenPlayground has recently announced that they are going to starting deleting accounts that access their API via automated means (probably as as a direct response to this library). You should be fine as long as you don't send to many requests, since this library is able to bypass their bot detection by spoofing the X-Session header. They've also recently been requiring SMS verification upon signup, but this can easily be bypassed by signing in using a Google account.

screenshot from their discord server

Features:

This library has the following abilities:

  • Log in using OTP code
  • List models
  • Generate text

Installation:

You can install this library by running the following command:

pip3 install openplayground-api

Documentation:

An example of how to use this library can be found in /examples/example.py.

The Model Class:

The openplayground.Model class describes a model that is available to the user. Valid attributes are:

  • provider - The company that developed the model (e.g., openai, anthropic)
  • name - The name of the model, such as text-davinci-003.
  • version - The version of the model. This may return None on some models.
  • tag - A string that combines the provider and name, such as openai:text-davinci-003.
  • params - A dictionary containing possible parameters for the model.

Authenticating With an OTP Code:

The openplayground.Auth class can be used to get your token using an OTP code emailed to you. Note that the following examples assume that auth is the name of your openplayground.Auth class.

import openplayground
auth = openplayground.Auth()

Sending the OTP Code:

The openplayground.Auth.send_otp_code function sends an email containing the OTP code to the specificed email address.

auth.send_otp_code("sample@example.com")

Verifiying the OTP Code:

Once you have the OTP code, you can use the openplayground.Auth.verify_otp_code function to get your token from that OTP code. You can then use this token to create an openplayground.Client instance.

otp_code = input("Enter OTP code: ")
token = auth.verify_otp_code()

Using the Client:

The openplayground.Client class accepts two arguments, which are your account's token and its email. Your token can be obtained from the __session field in your browser's cookies, or using the openplayground.Auth class as shown above. The email field is optional, but filling it out might reduce the chances of this library being detected.

import openplayground
client = openplayground.Client(token, email=email)

Note that the following examples assume client is the name of your openplayground.Client instance.

Downloading the Available Models:

The client.get_models function fetches the available models from https://nat.dev/api/all_models, and returns a dictionary of openplayground.Model objects. The client downloads the available models upon initialization and stores it in client.models, so calling this function shouldn't be necessary.

Some popular model tags are:

  • OpenAI: openai:gpt-4, openai:gpt-3.5-turbo, openai:text-davinci-003
  • Anthropic: anthropic:claude-instant-v1.0, anthropic:claude-v1.2
  • Facebook/Stanford: textgeneration:llama-65b, textgeneration:alpaca-7b
print(client.models.keys())
#dict_keys(['forefront:EleutherAI/GPT-J', 'forefront:EleutherAI/GPT-NeoX', 'forefront:pythia-12b', 'forefront:pythia-20b', 'forefront:pythia-6.9b', 'anthropic:claude-instant-v1.0', 'anthropic:claude-v1.2', 'textgeneration:alpaca-7b', 'textgeneration:llama-65b', 'huggingface:bigscience/bloomz', 'huggingface:google/flan-t5-xxl', 'huggingface:google/flan-ul2', 'cohere:command-medium-nightly', 'cohere:command-xlarge-nightly', 'cohere:medium', 'cohere:xlarge', 'openai:gpt-4', 'openai:code-cushman-001', 'openai:code-davinci-002', 'openai:gpt-3.5-turbo', 'openai:text-ada-001', 'openai:text-babbage-001', 'openai:text-curie-001', 'openai:text-davinci-002', 'openai:text-davinci-003'])

Generating Text:

The client.generate function generates some text given a model and a prompt. Optionally, you can also specify arguments such as the maximum length in the kwargs. You can find a list of valid arguments and their defaults in openplayground.Model.params. A few common ones are:

  • maximum_length
  • temperature
  • top_k
  • top_p

The values returned from this function are streamed and expressed in a dictionary. Note that GPT-4 access currently has a daily limit of around 10 requests/day, and may become paid in the future.

Streamed example:

for chunk in client.generate("openai:gpt-3.5-turbo", prompt):
  if chunk["event"] == "infer":
    print(chunk["message"], end="", flush=True)

Non-streamed example:

message = ""
for chunk in client.generate("openai:gpt-3.5-turbo", prompt):
  if chunk["event"] == "infer":
    message += chunk["message"]
print(message)

Misc:

Changing the Spoofed User-Agent:

You can change the global User-Agent by setting openplayground.user_agent right after importing the library.

import openplayground
openplayground.user_agent = "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/111.0.0.0 Safari/537.36"

Copyright:

This project is licensed under the GNU GPL v3. Most of the code has been written by me, ading2210. A list of all the contributors can be found here.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distributions

No source distribution files available for this release.See tutorial on generating distribution archives.

Built Distribution

openplayground_api-0.1.1-py3-none-any.whl (18.0 kB view details)

Uploaded Python 3

File details

Details for the file openplayground_api-0.1.1-py3-none-any.whl.

File metadata

File hashes

Hashes for openplayground_api-0.1.1-py3-none-any.whl
Algorithm Hash digest
SHA256 8f0eb04bda9ec5e9e4f521fade97c534feac5f4f4a51bd99ad23552e2f297ecd
MD5 f6c0dde9fa0f48006fc648d1029ac581
BLAKE2b-256 725094ce710750a66a01e92e301effd975d678dc9714397a81dfdf0a0ad9f0eb

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page