Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressive

Adoption judgment: this is worth considering only when the explicit goal is to study or try a low-refusal local model. It should not be adopted on the current evidence for production, compliance-sensitive workflows, predictable safety boundaries, reproducible evaluation, or official support. It is not a general hosted chatbot or an API product, and it is not presented here as an official Qwen release; a more accurate analogy is a community-made uncensored or jailbreak-oriented variant built around Qwen3.6-35B-A3B.

Its practical purpose is text generation while attempting to reduce refusals on requests that the base model might block. One X post claims “0/465 refusals,” “fully unlocked,” and “no capability loss,” while another post mentions local deployment. These are social-media claims rather than independently documented results. The available evidence includes no benchmark, test protocol, prompt set, or reproducible comparison, so it supports the model’s positioning and advertised behavior more than it supports consistent performance across tasks, languages, or prompts.