4 ms·

It is shared between users and better utilized and optimized.

• otabdeveloper4 2y ago
  "Sharing between users" doesn't make it cheaper. It makes it more expensive due to the inherent inefficiencies of switching user contexts. (Unless your sales people are doing some underhanded market segmentation trickery, of course.)
  • manmal 2y ago
    No, batched inference can work very well. Depending on architecture, you can get 100x or even more tokens out of the system if you feed it multiple requests in parallel.
    • api 2y ago
      Couldn't you do this locally just the same? Of course that doesn't map well to an individual chatting with a chat bot. It does map well to something like "hey, laptop, summarize these 10,000 documents."
      • manmal 2y ago
        Yes, and people do that. Some people get thousands of tokens per second that way, with affordable setups (eg 4x 3090). I was addressing GP who said there is no economies of scale to having multiple users.