3 ms·
I am not sure how much bigger, but definitely much bigger IMHO. Otherwise you wouldn't be capped at 25 requests every 3h. That number is small enough that makes
by marcyb5st 4y ago
I am not sure how much bigger, but definitely much bigger IMHO. Otherwise you wouldn't be capped at 25 requests every 3h. That number is small enough that makes me think the inference costs/hardware needed are much bigger than 3.5.
- pixl97 4y agoI believe I heard that running inference longer is giving the better responses we're seeing in v4. Hence why v4 is taking so much longer to output data. Of course we won't know this for sure until OAI tells us, so we may be in the dark for a while.