ai-api-experiments

Table of Contents

• Summary
• Reference
• Features
• Getting started

Summary

Applications providing a user-friendly interface to interact with OpenAI's AIaaS web-backends

Reference

• OpenAI API Reference
• OpenAI Models
• Get started with the Gemini API

Features

• Chat bot CLI Tool utilizing gpt-3.5-turbo or gpt-4-turbo model
• Image generator CLI Tool utilizing dall-e-2 or dall-e-3 model
• Speech generator CLI Tool utilizing tts-1 or tts-1-hd model
• Desktop application designed to provide a user-friendly interface for interacting with OpenAI’s AIaaS web-backends enabling image generation, speech generation and chatting functionality

Getting started

Preconditions

• If your IDE supports it, install Dev Containers extension or download a Python >=3.9.x Release
• Install pip dependencies: pip install -r requirements.txt
• Create an API key in your OpenAI organization
• Export the API key in a terminal process: export OPENAI_API_KEY="<your OPENAI_API_KEY>"

Utilizing CLI tools

Chat bot

Run:

python -m cli.chat_bot_app --prompt "<your prompt>" --model "<gpt-4o, gpt-4o-mini, o1, o3-mini>"

On larger prompts consider creating a separate file and reading the content of the file into a variable:

vi prompt.txt
prompt=$(cat prompt.txt)
python -m cli.chat_bot_app --prompt "$prompt" --model "<gpt-4o, gpt-4o-mini, o1, o3-mini>"

NOTE: You might need to change your billing plan, if you encounter the error openai.RateLimitError: Error code: 429 - {'error': {'message': 'You exceeded your current quota, please check your plan and billing details. For more information on this error, read the docs: https://platform.openai.com/docs/guides/error-codes/api-errors.', 'type': 'insufficient_quota', 'param': None, 'code': 'insufficient_quota'}}

Image generator

Install pip dependencies and run application:

python -m cli.image_generator_app --model "<dall-e-2, dall-e-3>" --prompt "<your prompt>" --size "<size of the image to be generated, e.g. 1024x1024>"
# Checkout link of `Generated image URL` in a browser of choice

NOTE: "When using DALL·E 3, images can have a size of 1024x1024, 1024x1792 or 1792x1024 pixels."

Speech generator

Install pip dependencies and run application:

python -m cli.speech_generator_app --model "tts-1-hd" --voice "alloy" --prompt "<your prompt>"

Launching Desktop app

Run:

python -m desktop.companion

Resulting images: