zigwangles/Huihui-Qwen3-Coder-30B-A3B-Instruct-abliterated

TEXT GENERATIONPricing:Input $0.5 / Output $1.8Concurrent Unit Cost:1Model Size:30BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Sep 14, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

The zigwangles/Huihui-Qwen3-Coder-30B-A3B-Instruct-abliterated model is a 30 billion parameter instruction-tuned causal language model based on the Qwen3-Coder architecture. Developed by huihui-ai, this version has undergone an "abliteration" process to significantly reduce safety filtering and uncensor its responses. It is primarily intended for research and experimental use where the removal of refusal behaviors is desired, particularly in code generation contexts.

Loading preview...

Model Overview

This model, Huihui-Qwen3-Coder-30B-A3B-Instruct-abliterated, is a 30 billion parameter instruction-tuned variant of the Qwen3-Coder architecture. Its key differentiator is the application of an "abliteration" process, a method designed to remove refusal behaviors and uncensor the model's outputs. This makes it a proof-of-concept for exploring LLM responses without typical safety constraints.

Key Characteristics

  • Uncensored Output: Safety filtering has been significantly reduced, allowing for potentially sensitive or controversial content generation.
  • Abliteration Process: Utilizes a novel and faster abliteration method, yielding improved results in refusal removal compared to previous approaches.
  • Code Generation Base: Built upon the Qwen3-Coder architecture, suggesting underlying capabilities in code-related tasks.
  • Ollama Integration: Directly available for use with Ollama via huihui_ai/qwen3-coder-abliterated.

Usage Warnings & Considerations

Due to the removal of safety filters, users must be aware of several critical points:

  • Risk of Inappropriate Content: The model may generate sensitive, controversial, or inappropriate outputs.
  • Not for All Audiences: Unsuitable for public-facing applications, underage users, or environments requiring strict content moderation.
  • Legal and Ethical Responsibility: Users are solely responsible for ensuring their usage complies with all applicable laws and ethical standards.
  • Research Focus: Recommended for research, testing, or controlled experimental environments rather than production use.
  • No Safety Guarantees: huihui.ai explicitly states no responsibility for consequences arising from its use, as it lacks rigorous safety optimization.

Finetunes
1 models