Tokens are returned from the chat completions API in big, choppy batches. They used to be smoothly returned one by one. See this example import time from openai import OpenAI client = OpenAI() response = client.chat.completions.create( model="gpt-4o", messages=[{"role": "user", "content"…

Thank you so much for reporting the issue! We’ve done some investigation, and this was caused by a configuration change on our end, which we have now reverted. We also recommend using our official library for streaming as robust client code can help protect against this. Sorry for the trouble, and…

Streaming events returned bunched up

API Bugs

PaulBellow July 17, 2024, 9:08pm 3

Might be related to this…

Do you mind if I merge the threads?

Topic		Replies	Views
We need source links and we need them now API	33	11043	December 24, 2023
Availability of OpenAI api doc dump Documentation openai-documentation	4	1658	November 17, 2023
I believe I am the Timelord Community	42	2185	December 25, 2023
Is API completion model davinci-002 and its 16385 context useful for ANYTHING? API davinci	1	2409	December 14, 2023
How do you maintain historical context in repeat API calls? API	29	79274	December 23, 2023

Streaming events returned bunched up

Related Topics