Assistants API is unusable on prod

It is taking 9 seconds and up to complete a request, the one below took 22 seconds
this is the content
What are the DUI penalties for a second DUI in New York?
I am polling for the below messages every second
queued
in_progress
in_progress
in_progress
in_progress
in_progress
in_progress
in_progress
in_progress
in_progress
in_progress
in_progress
in_progress
in_progress
in_progress
in_progress
in_progress
in_progress
in_progress
in_progress
in_progress
completed
This is the messages

What model is your Assistant using? Have you tried this with GPT 3.5 (out of interest)?

Use the Chat API whenever possible.
The Assistant API doesn’t support streaming, which makes it look “slow”, and you have to wait for the message to finish generating. Also, if it needs to fetch uploaded documents, it might take a extra couple of seconds.
There’s really no “solution” to that. GPT4 is rather “slow”. If possible, try to use 3.5, as it’s faster (but also not as “intelligent”)