As far as I know this is not a thinking model. But after a long conversation with qwen 3.5 27b, I switched the conversation over to qwen3 coder next to handle the longer context, and it started thinking? I 100% verfied that this was running on my llama.cpp server as 'unsloth/Qwen3-Coder-Next-GGUF:UD-Q8_K_XL' utilizing my GPU. Im baffled and confused at what is happening here.