q66117593/Huihui-Qwen3.6-35B-A3B-Claude-4.7-Opus-abliterated

Hugging Face
TEXT GENERATIONConcurrent Unit Cost:3Model Size:35.1BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Aug 13, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Warm

The q66117593/Huihui-Qwen3.6-35B-A3B-Claude-4.7-Opus-abliterated model is a 35.1 billion parameter language model based on the Qwen3.6 architecture, developed by huihui-ai. This model is an uncensored variant of lordx64/Qwen3.6-35B-A3B-Claude-4.7-Opus-Reasoning-Distilled, created using an 'abliteration' technique to remove safety refusals. It is specifically designed for research and experimental use in environments where reduced safety filtering is desired, allowing for potentially sensitive or controversial outputs.

Loading preview...

Model Overview

The q66117593/Huihui-Qwen3.6-35B-A3B-Claude-4.7-Opus-abliterated is a 35.1 billion parameter language model derived from the Qwen3.6 architecture. Developed by huihui-ai, this model is a modified version of lordx64/Qwen3.6-35B-A3B-Claude-4.7-Opus-Reasoning-Distilled with its safety filtering significantly reduced through an 'abliteration' process. This modification aims to enable the generation of content that might typically be refused by standard models.

Key Characteristics

  • Uncensored Output: Safety filtering has been substantially reduced, allowing for a wider range of generated content, including potentially sensitive or controversial topics.
  • Experimental Nature: Created as a proof-of-concept for removing refusals from LLMs without using TransformerLens.
  • Ollama Integration: Directly available for use with Ollama, simplifying local deployment.

Usage Warnings and Recommendations

Due to its reduced safety mechanisms, users should be aware of several critical points:

  • Risk of Inappropriate Content: The model may generate sensitive, controversial, or inappropriate outputs.
  • Not for All Audiences: Outputs may be unsuitable for public settings, underage users, or applications requiring high security.
  • User Responsibility: Users are solely responsible for ensuring compliance with legal and ethical standards for generated content.
  • Recommended Use: Best suited for research, testing, or controlled environments, rather than production or public-facing commercial applications.
  • Monitoring Advised: Real-time monitoring and manual review of outputs are strongly recommended.

Finetunes
1 models