async-openai

Unofficial Async Python client library for the OpenAI API based on Documented Specs

Features

• Asyncio based with Sync and Async Support with httpx

• Supports all API endpoints

  • Completions

  • Edits

  • Embeddings

  • [Images] (https://beta.openai.com/docs/api-reference/images)

  • Files

  • Fine-Tunes

  • Moderations

  • [Search]

• Strongly typed validation of requests and responses with Pydantic Models with transparent access to the raw response and object-based results.

• Handles Retries automatically through backoff

• Enables Remote Cloud Object Storage File Handling through file-io

Installation

# Install from stable
pip install async-openai

# Install from dev/latest
pip install git+https://github.com/GrowthEngineAI/async-openai.git

Quick Usage

import asyncio
from async_openai import OpenAI, settings

# Environment variables should pick up the defaults
# however, you can also set them explicitly.

# `api_key` - Your OpenAI API key.  Env: [`OPENAI_API_KEY`]
# `api_base` - The base URL of the OpenAI API. Env: [`OPENAI_API_BASE`]
# `api_type` - The OpenAI API type.  Env: [`OPENAI_API_TYPE`]
# `api_version` - The OpenAI API version.  Env: [`OPENAI_API_VERSION`]
# `organization` - The OpenAI organization. Env: [`OPENAI_ORGANIZATION`]
# `proxies` - A dictionary of proxies to be used. Env: [`OPENAI_PROXIES`]
# `timeout_secs` - The timeout in seconds to be used. Env: [`OPENAI_TIMEOUT_SECS`]
# `max_retries` - The number of retries to be used. Env: [`OPENAI_MAX_RETRIES`]

settings.configure(
    api_key = "sk-XXXX",
    organization = "org-XXXX",
)

# [Sync] create a completion
# Results return a CompletionResult object
result = OpenAI.completions.create(
    prompt = 'say this is a test',
    max_tokens = 4,
    stream = True
)

# print the completion text
# which are concatenated together from the result['choices'][n]['text']

print(result.completion_text)

# print the number of choices returned
print(len(result))

# [Async] create a completion
# All async methods are generally prefixed with `async_`

result = asyncio.run(
    OpenAI.completions.async_create(
        prompt = 'say this is a test',
        max_tokens = 4,
        stream = True
    )
)

Alternative Usage

Following a similar pattern to the original openai python project, you can additionally use the methods as a drop-in replacement for most functions

import asyncio
import async_openapi as openai

# Environment variables should pick up the defaults
# however, you can also set them explicitly.

openai.settings.configure(
    api_key = "sk-XXXX",
    organization = "org-XXXX",
)

# [Sync] create a completion
# Results return a CompletionResult object
# Note, the only difference in this method is that the `Completion` is capitalized to match the OpenAI API

result = openai.Completions.create(
    prompt = 'say this is a test',
    max_tokens = 4,
    stream = True
)

# print the completion text
# which are concatenated together from the result['choices'][n]['text']

print(result.completion_text)

# print the number of choices returned
print(len(result))

# [Async] create a completion
result = asyncio.run(
    openai.Completions.async_create(
        prompt = 'say this is a test',
        max_tokens = 4,
        stream = True
    )
)

Dependencies

The aim of this library is to be as lightweight as possible. It is built on top of the following libraries:

• httpx: Async / Sync HTTP Requests

• pydantic: Type Support

• loguru: Logging

• file-io: Async Cloud-based File Storage I/O

• backoff: Retries with Exponential Backoff