How can I fix the AttributeError when using openai.ChatCompletion?

Below is a brief explanation and an updated example:

What’s Happening

• AttributeError:
  The error “module ‘openai’ has no attribute ‘ChatCompletion’” typically happens if your installed OpenAI Python package is outdated. New endpoints like ChatCompletion were added in recent updates.

• Incorrect Endpoint Use:
  The snippet you posted uses:

  PYTHONCopied
  response = openai.ChatCompletion.create(
      engine="text-davinci-003",
      prompt=query_text
  )

  Two issues exist here:

  1. The model name "text-davinci-003" is meant for the text completions endpoint.
  2. The ChatCompletion endpoint expects a list of messages (using a “role” and “content”), and the parameter is named model (not engine).

How to Fix It

1. Update Your Package:
   Make sure you have the latest version:

   BASHCopied
   pip install --upgrade openai

2. Choose the Correct Endpoint:

   • For Chat Completions:
     Use a request that provides messages. For example:

     PYTHONCopied
     import openai
     
     response = openai.ChatCompletion.create(
         model="gpt-4o-mini",
         messages=[
             {"role": "user", "content": "Say this is a test"}
         ]
     )
     print(response.choices[0].message.content)

   • For Text Completions:
     If you want to use "text-davinci-003", use the Completion endpoint:

     PYTHONCopied
     import openai
     
     response = openai.Completion.create(
         engine="text-davinci-003",
         prompt="Your prompt text here",
         max_tokens=50
     )
     print(response.choices[0].text)

Summary

• Update your openai package.
• Use openai.ChatCompletion.create with the correct parameters (i.e. model and a list of messages) for chat-based completions.
• Use openai.Completion.create with engine and prompt if you want to work with text completions using text-davinci-003.

This adjustment should address the error and help get your API calls working properly.