LocalLLaMA

Qwen3.6 27B seems struggling at 90k on 128k ctx windows

I have RX 7900 XTX, running Qwen3.6 27B Q4_K_XL. got 400ish pp and 30s tps. every work below 64k is incredible and it spits out good quality code. But i tried to push it further to work on kinda complex devops related work and it fail at tool calling a…