Elektrine lite

← Feed

@fasterandworse@hci.social

2026-02-20 22:13 UTC

@bjn adding a note for that haha

Replies (1)

  • @bjn@mstdn.social 2026-02-20 22:18

    @fasterandworse Yep. If you aren’t paying OpenAI for inference, how are they going to cover training costs? I have a feeling that LLMs will become more tightly focused on a domain and so easier to train, at which point they might commodity items.

    Open ##1544796