• Razen@lemmy.world
    link
    fedilink
    English
    arrow-up
    16
    ·
    2 days ago

    Are they eating the cost? How are they able to do it while others are unable to?

    • Balinares@pawb.social
      link
      fedilink
      English
      arrow-up
      18
      ·
      2 days ago

      They invented a hybrid attention design that drastically reduces the amount of memory needed for the KV cache at inference time. Like, dividing it by 10. And memory is a large part of the cost of inference.

      • jaykrown@lemmy.worldOP
        link
        fedilink
        English
        arrow-up
        8
        ·
        2 days ago

        This is a main part of the reason, yes. They actually innovated and did something that pushed the technology forward to be much more efficient, which we first saw with DeepSeek R1 for different reasons.

          • WhyJiffie@sh.itjust.works
            link
            fedilink
            English
            arrow-up
            1
            ·
            10 hours ago

            probably even with duck.ai. nobody could prove they are doing it, even if they are doing it

            I don’t use any of tiktok or meta but I believe tiktok is more addictive by design.

        • Wispy2891@lemmy.world
          link
          fedilink
          English
          arrow-up
          5
          arrow-down
          1
          ·
          2 days ago

          i never said that. In fact, openai gives for free 2.5 million of gpt-5.5 tokens every day if you share all your inputs for training.

          It was the answer to “why it’s this cheap?” => because it’s subsidized by your data

          • magic_smoke@lemmy.blahaj.zone
            link
            fedilink
            English
            arrow-up
            4
            ·
            1 day ago

            I think what they meant is “why is this so cheap relative to American competitors since they’re all stealing your data just as much.”