DavidAU/Qwen3.5-2B-Claude-4.6-OS-Auto-Variable-HERETIC-UNCENSORED-THINKING

VISIONConcurrent Unit Cost:1Model Size:2.3BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Mar 10, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

DavidAU/Qwen3.5-2B-Claude-4.6-OS-Auto-Variable-HERETIC-UNCENSORED-THINKING is a 2.3 billion parameter Qwen 3.5 dense model fine-tuned by DavidAU using four Claude-4.6-OS datasets. This model is designed to be fully uncensored and 'heretic,' meaning it operates without safety alignments and follows user instructions without questioning. It excels in reasoning and output generation, surpassing the base Qwen 3.5 2B model on various benchmarks, and supports multimodal inputs including vision.

Loading preview...

Model Overview

This model, DavidAU/Qwen3.5-2B-Claude-4.6-OS-Auto-Variable-HERETIC-UNCENSORED-THINKING, is a 2.3 billion parameter Qwen 3.5 dense model fine-tuned by DavidAU. It leverages four Claude-4.6-OS datasets to enhance reasoning and output generation, outperforming the base Qwen 3.5 2B model on most benchmarks. A key differentiator is its 'heretic' and fully uncensored nature, designed to execute user requests without safety alignments.

Key Capabilities

  • Uncensored Output: Provides responses without built-in safety restrictions.
  • Enhanced Reasoning: Improved reasoning capabilities due to Claude dataset fine-tuning.
  • Multimodal Support: Capable of processing vision inputs, with video understanding also supported (though video portions were not tested by the developer).
  • Benchmark Performance: Shows improved scores over the base Qwen3.5-2B model on various language and vision benchmarks, including ARC, BoolQ, and PIQA.

Good For

  • Creative Writing & Roleplay: Its uncensored nature and strong output generation make it suitable for unrestricted creative tasks.
  • Complex Reasoning Tasks: Benefits from Claude-like reasoning, often incorporating 'thinking' blocks in responses.
  • Applications Requiring Unfiltered Responses: Ideal for use cases where strict adherence to user prompts, regardless of content, is paramount.
  • Local Deployment: Optimized for local hardware, with recommendations for q4ks or IQ3S quantization for efficient operation.

Finetunes
1 models