• VOIDPTR@lemmy.ml
    link
    fedilink
    arrow-up
    8
    ·
    10 days ago

    I have been using DeepSeek, MiMo, GLM and other Chinese models for 6 months now and don’t see the reason to ever pay for US frontier models.

    • sudoer777@lemmy.ml
      link
      fedilink
      English
      arrow-up
      3
      ·
      10 days ago

      So far I’ve never seen DeepSeek V4.1 Flash really choke on anything I’ve handed it

  • Alex@lemmy.ml
    link
    fedilink
    arrow-up
    7
    ·
    10 days ago

    I do wonder what their funding strategy is? Is it just hosting the bigger foundation models? To me the most interesting aspect of the open models is taking the weights and creating specialised versions you can run locally on less compute.