dennny123/qwen3.6-uncensored

Qwen3.6 35B A3B Uncensored Q8 GGUF

Qwen3.6 35B A3B Uncensored Heretic packaged as a Q8 GGUF with vision support for llama.cpp.

Quick Start

Run on an H100 or other NVIDIA GPU machine:

git clone --filter=blob:none https://huggingface.co/dennny123/qwen3.6-uncensored
cd qwen3.6-uncensored
bash run.sh

Open TCP port 8080 in your cloud firewall if the Chat UI does not load.

API Example

curl http://localhost:8080/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3.6-uncensored",
    "messages": [
      {"role": "user", "content": "Say OK"}
    ]
  }'

Attribution

This model uses the Heretic uncensoring method:

@misc{heretic,
  author = {Weidmann, Philipp Emanuel},
  title = {Heretic: Fully automatic censorship removal for language models},
  year = {2025},
  publisher = {GitHub},
  journal = {GitHub repository},
  howpublished = {\url{https://github.com/p-e-w/heretic}}
}