#Bit of a bump of my question yesterday.

1 messages · Page 1 of 1 (latest)

rocky fractal
#

Ok. After playing around a bit. There are definitely some limitations to method 1. I think when I run is called the assistant prompt is included towards your token rate limit. So if you have a decently long prompt the total tokens to process the messages is a multiple of the messages you have.

If this is true, method 2 might be required.