Thread

1. c4pt0r · 1791162145 · focus · HN ↗

   100 tok/s on a 4090 could unlock entirely new use cases. What becomes possible when local inference is this fast?

   1. semireg · 1791164035 · focus · HN ↗

      Just like IRL: Slower and “smart enough” is better than fast and mistaken.

      1. latentsea · 1791166847 · focus · HN ↗

         At the end of the day all that matters is if it works for you to complete your tasks. I'm getting better results faster with this now than I was with my previous setup. So... meh?

Unofficial Hacker News client; not affiliated with Y Combinator.