How to correctly pass prompt instructions using chat.completions.create with OpenAI Python library 1.8.0

It looks like you're mixing parameters from different endpoints. The chat completions endpoint expects a list of message objects (using the messages parameter) and no longer accepts a standalone prompt parameter.

For chat completions, you should include your instruction within the messages list. For example:

PYTHONCopied
client = OpenAI(api_key="xxx")
response = client.chat.completions.create(
    model="gpt-3.5-turbo",
    messages=[
        {"role": "system", "content": "Be short and precise"},
        {"role": "user", "content": "Your actual input message here."}
    ],
    temperature=0,
    max_tokens=1000
)

In this setup, the instruction ("Be short and precise") is provided in a system message. Adjust your messages list as needed. Removing the prompt parameter should resolve the exception.