Skip to content

Connect to OpenAI from Python

Connect to OpenAI from Python

This guide shows the minimal production-appropriate setup for calling OpenAI from Python:

install the SDK
store the API key safely
make a first request
keep the example easy to adapt

1. Install the SDK

python -m pip install openai python-dotenv

2. Store the API Key

Set the key as an environment variable.

Windows PowerShell

$env:OPENAI_API_KEY="your-api-key-here"

macOS or Linux

export OPENAI_API_KEY="your-api-key-here"

You can also store it in a local .env file for development:

OPENAI_API_KEY=your-api-key-here

3. Make a First Request

from openai import OpenAI

client = OpenAI()

response = client.responses.create(
    model="gpt-5.2",
    input="Explain retrieval-augmented generation in two short sentences."
)

print(response.output_text)

Why This Example Uses `responses.create`

The current OpenAI docs position the Responses API as the main path for new text-generation work. If you are copying older tutorials, you may still see chat-completions examples. Those older examples are useful historical context, but this page is intentionally using the newer API shape.

4. Add `.env` Support for Local Development

import os
from dotenv import load_dotenv
from openai import OpenAI

load_dotenv()

client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))

Common Mistakes

hardcoding the key in source files
using the wrong environment variable name
assuming older SDK examples still reflect the preferred API shape
coupling your app too tightly to one specific model name

Practical Advice

keep the API key in the environment, not in code
move model names into config once the app grows
start with one small test script before integrating into a larger app