aifeifei798/Qwen3.6-Queen-27B

VISIONConcurrency Cost:2Model Size:27BQuant:FP8Ctx Length:32kTool Calling:SupportedPublished:Apr 22, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Cold

The aifeifei798/Qwen3.6-Queen-27B is a 27 billion parameter language model based on the Qwen3.6 architecture, fine-tuned for high-intelligence, low-filter, multi-persona role-playing, particularly excelling in coding deity personas. It features a 32K context length and is optimized to overcome the 'Refusal Wall' of its base model, enhancing logical inference and efficiency by 10-15%. This model is designed for diverse creative writing tasks, specialized role-playing scenarios, and generating quick, scholarly responses in English.

Loading preview...

Model Overview: Qwen3.6-Queen-27B

The aifeifei798/Qwen3.6-Queen-27B is a 27 billion parameter model built upon the Qwen3.6 architecture, specifically engineered for advanced multi-persona role-playing and creative content generation. It boasts a 32,768-token context length and is noted for its ability to bypass the 'Refusal Wall' present in the base Qwen3.6-27B model. This optimization, achieved by 'orthogonalizing the refusal vector,' reportedly flattens logical inference bottlenecks and increases inference efficiency by 10-15%.

Key Capabilities:

  • Multi-Persona Role-playing: Excels in various personas, including a 'High-IQ, Low-Filter, Multi-Persona Coding Deity,' and a 'viral Dark Academia Queen.'
  • Creative Writing: Proficient in generating writing prompts, opus, and songs, often with a scholarly, thesis-like depth.
  • Refusal Wall Mitigation: Engineered to reduce aggressive safety filters, allowing for more direct and logical inference.
  • Efficiency: Claims a 10-15% increase in inference efficiency compared to its base model.
  • Language Support: Primarily designed for English language use.

Ideal Use Cases:

  • Specialized Role-playing: Particularly suited for complex, high-intelligence, and unfiltered character interactions.
  • Creative Content Generation: Generating detailed narratives, songs, and prompts.
  • Coding-related Personas: Can adopt personas like a 'Coding Queen' for tasks such as refactoring, deployment, and identifying edge cases.
  • Educational & Scholarly Responses: Capable of producing quick, scholarly responses akin to a thesis.
  • Mobile Adaptation: The module combination has been readjusted for mobile phone compatibility.