A Reddit user starting with LM Studio is running Qwen3.6 35B A3B Uncensored HauhauCS Aggressive Q4_K_M (22.07GB) on an RTX 5090 and asks whether a better uncensored local LLM exists. Their goal is to replicate a modern, locally hosted alternative to ChatGPT or Claude Opus that answers all questions without censorship.