Roman0/Qwen3-0.6B-heretic

Hugging Face
TEXT GENERATIONPricing:Input $0.04 / Cached $0.008 / Output $0.08Concurrent Unit Cost:1Model Size:0.8BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Dec 12, 2025License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Warm

Roman0/Qwen3-0.6B-heretic is a 0.8 billion parameter causal language model, based on the Qwen3-0.6B architecture, specifically modified to be uncensored using the Heretic v1.1.0 process. This model retains the Qwen3's 32,768 token context length and its unique ability to switch between thinking and non-thinking modes, while significantly reducing refusal rates compared to the original. It is optimized for use cases requiring less restrictive content generation and advanced reasoning capabilities.

Loading preview...

Roman0/Qwen3-0.6B-heretic: An Uncensored Qwen3 Variant

This model is a decensored version of Qwen/Qwen3-0.6B, created using the Heretic v1.1.0 tool. It maintains the core capabilities of the Qwen3 series, including its 0.6 billion parameters and a substantial 32,768 token context length, while being specifically modified to reduce content refusals.

Key Differentiators & Capabilities

  • Decensored Output: Significantly reduces content refusals, with a reported 3 refusals out of 100, compared to 57/100 in the original Qwen3-0.6B model.
  • Flexible Thinking Modes: Inherits Qwen3's unique ability to seamlessly switch between a 'thinking mode' for complex logical reasoning, math, and coding, and a 'non-thinking mode' for efficient, general-purpose dialogue. This can be controlled via enable_thinking parameter or /think and /no_think tags in prompts.
  • Enhanced Reasoning: Benefits from Qwen3's advancements in reasoning, instruction-following, and agent capabilities, particularly in its thinking mode.
  • Multilingual Support: Supports over 100 languages and dialects, offering strong multilingual instruction following and translation.
  • Agentic Use: Excels in tool-calling capabilities, recommended for use with Qwen-Agent for complex agent-based tasks.

When to Use This Model

This model is particularly suited for applications where a less restrictive content policy is desired, combined with the advanced reasoning and multilingual capabilities of the Qwen3 architecture. It is ideal for developers seeking a compact model that can handle complex logical tasks and creative writing without the content filtering present in the base model.

Finetunes
1 models