llama.cpp

0a2420a4 - models: support vllm decision 2 qwen 3.5 variant