Comment by computerex

You can use my new golang inference engine to run variants of Qwen 3.5 faster than llama.cpp: https://github.com/computerex/dlgo