It is small but still really good

  • Xylight@lemdro.id
    link
    fedilink
    English
    arrow-up
    2
    ·
    3 days ago

    You can run Qwen3 4b thinking at q4 quantization at 2.5GB, which is probably a better model too