Yumina

11 replies

  1. kljws · Staff ·

    fable5.2(并非)

  2. Vyh ·

    GLM

  3. 大大爆肝 ·

    我也用这个了

  4. 大大爆肝 ·

    啊

  5. Adfvt ·

    The more context lenght a model has the better

  6. Oncin ·

    hmmm deepseek v4.1 flash

  7. Adam ·

    Why deepseek v4.1 flash

  8. Oncin ·

    the models is better at thinking in "object" like gemini flash models just weaker because its cheap

  9. 大大爆肝 ·

    这个模型比GLM5.3好很多吗?因为DK4.1最高消耗和默认模型的gemini3.1好像差不多啊

  10. Oncin ·

    i don't know about GLN5.3 i never used it, it's just that in my exprience even though deepseek v4.1 it's a weak model, i think the reasoning and its underlying model is very similar to gemini flash 3.7 and 3.8, it's not as powerful as gemini flash though

  11. Oncin ·

    maybe i'm biased, but it's just printing out better result comparing to other cheap model like gpt 6 luna, claude haiku etc