GPT-4.5 API Tutorial: Getting Started With OpenAI's API

This tutorial guides you through using OpenAI's GPT-4.5 language model via its API, focusing on building a Python-based chatbot. This offers a cost-effective alternative to the monthly ChatGPT subscription.

Understanding GPT-4.5

GPT-4.5, OpenAI's latest model, excels in conversational fluency, contextual understanding, and factual accuracy. It generates more natural responses and adapts tone effectively, making it ideal for chatbots and content creation. However, unlike OpenAI's O-series models, it lacks step-by-step reasoning capabilities, so it's less suitable for complex problem-solving tasks.

Connecting to the OpenAI API

This involves obtaining an API key and setting up your Python environment.

1. Obtaining Your API Key:

• Visit the OpenAI API key page.
• Sign in or create an OpenAI account.
• Click "Create new secret key." Crucially, copy this key immediately; you won't be able to retrieve it later.

• Store the key securely in a .env file (in the same directory as your Python script) using the format: OPENAI_API_KEY=<your_api_key></your_api_key>

2. API Pricing:

OpenAI's API is pay-per-use, charging by token (approximately ¾ of a word). This is often cheaper than a subscription for regular use. GPT-4.5, the model used in this tutorial, is OpenAI's most advanced general-purpose model.

3. Setting Up Your Python Environment:

Use Anaconda to create a clean Python environment:

conda create -n gpt45 -y python=3.9
conda activate gpt45
pip install openai python-dotenv

4. Making Your First API Request:

Create a file named script.py and add the following code:

from openai import OpenAI
from dotenv import load_dotenv
import os

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

completion = client.chat.completions.create(
    model="gpt-4.5-preview",
    messages=[{"role": "user", "content": "Hello"}],
)
print(completion.choices[0].message.content)

Run this using python script.py.

Building a GPT-4.5 Chatbot:

Enhance the script to create an interactive chatbot:

from openai import OpenAI
from dotenv import load_dotenv
import os

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

chat_history = []
while True:
    prompt = input("> ")
    if prompt == "exit":
        break
    chat_history.append({"role": "user", "content": prompt})
    completion = client.chat.completions.create(
        model="gpt-4.5-preview", messages=chat_history
    )
    answer = completion.choices[0].message.content
    print(answer)
    chat_history.append({"role": "assistant", "content": answer})

This chatbot maintains conversation history, allowing for more contextually aware responses.

Conclusion:

This tutorial demonstrated how to leverage the OpenAI API and GPT-4.5 to build a functional chatbot using Python. Remember to handle your API key securely and be mindful of API usage costs.

The above is the detailed content of GPT-4.5 API Tutorial: Getting Started With OpenAI's API. For more information, please follow other related articles on the PHP Chinese website!