A fine-tuned variant of Qwen3's 8B architecture scaled to 27B parameters, this model is oriented toward NSFW content generation based on its explicit labeling. It accepts both text and image inputs, giving it multimodal awareness. As a community-published fine-tune, behavior and safety guardrails differ significantly from the base Qwen3 model.