8 ms·I'm using mixtral-8x7b-v0.1.Q4_K_M.gguf with llama.cpp and it only requires 25GB.by ranger_danger 3y agoI'm using mixtral-8x7b-v0.1.Q4_K_M.gguf with llama.cpp and it only requires 25GB.