CCSSNE/trohrbaugh-Qwen3.6-27B-heretic

VISIONConcurrency Cost:2Model Size:27BQuant:FP8Ctx Length:32kTool Calling:SupportedPublished:May 22, 2026License:apache-2.0Architecture:Transformer Open Weights Cold

CCSSNE/trohrbaugh-Qwen3.6-27B-heretic is a 27 billion parameter causal language model, a decensored version of Qwen/Qwen3.6-27B, created using Heretic v1.2.0+custom. It features a 262,144 token context length, extensible up to 1,010,000 tokens with YaRN scaling, and is optimized for agentic coding, handling frontend workflows and repository-level reasoning with enhanced fluency and precision. This model also introduces 'Thinking Preservation' to retain reasoning context from historical messages, streamlining iterative development and reducing overhead.

Loading preview...

Model Overview

CCSSNE/trohrbaugh-Qwen3.6-27B-heretic is a 27 billion parameter causal language model, a decensored variant of the original Qwen/Qwen3.6-27B, created using Heretic v1.2.0+custom. It maintains the original model's robust architecture, including a native context length of 262,144 tokens, extensible up to 1,010,000 tokens through YaRN scaling techniques. A key differentiator of this specific model is its significantly reduced refusal rate (9/100 compared to 99/100 for the original), indicating a less restrictive output policy.

Key Capabilities

  • Decensored Output: Achieves a refusal rate of 9/100, offering more permissive content generation compared to the base model.
  • Agentic Coding: Excels in complex coding tasks, including frontend workflows and repository-level reasoning, with improved fluency and precision.
  • Thinking Preservation: Features a unique option to retain reasoning context from historical messages, enhancing iterative development and decision consistency.
  • Multimodal Support: Capable of processing text, image, and video inputs, making it suitable for diverse applications.
  • Ultra-Long Context: Natively supports 262,144 tokens, extendable to over 1 million tokens with YaRN, beneficial for long-horizon tasks.

Good For

  • Developers requiring less restrictive content generation: Ideal for use cases where the base model's refusal policies are too stringent.
  • Advanced Coding Agents: Particularly strong in agentic coding scenarios, handling complex development tasks and maintaining reasoning context.
  • Multimodal Applications: Suitable for tasks involving image and video understanding, alongside text processing.
  • Long-Context Tasks: Excellent for applications needing to process and generate very long texts, such as document analysis or extensive codebases.

Finetunes
4 models