llama.cpp

model : fix Qwen3.5 model type detection

#20126

Merged