Llama.cpp serves Qwen3.6-35B-A3B on RTX 4070 Ti Super

The author demonstrates how to turn an underutilized gaming PC equipped with an RTX 4070 Ti Super into a local AI server using llama.cpp and a Qwen MoE model. By running a single command with specific flags, the setup streams model output to multiple household devices over Wi-Fi with minimal performance loss compared to the host. Open article →