The API currently doesn’t support streaming completions with n>1. When I run the following code
import os import openai openai.api_key = os.getenv("OPENAI_API_KEY") stream = openai.Completion.create( engine="text-davinci-001", prompt="", max_tokens=100, n=3, temperature=1, stream=True )
I get the exact same output for all three completions. Is this a bug? Is it possible to support this use-case in the near future?