A

AliceAI-Foundation-80B-A3B

M5 Max 128GB · Q4_K_M · llama.cpp

M5 Max 128GBで約65 tok/s、45.10GiB、swap増加なし。

約65. MLX Q4 on the same box was quoted 46–48; rank uses the GGUF print.

Officialclaimed

@aqtyGet