LucidCC13105/humanizer

TEXT GENERATIONPricing:Input $1.2 / Cached $0.24 / Output $4.8Concurrent Unit Cost:1Model Size:12BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Oct 17, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

LucidCC13105/humanizer is a 12 billion parameter model designed to rewrite AI-generated text to sound more human. Built upon the Google Gemma-4-12B architecture, it specializes in transforming drafts like emails, essays, and reports in both English and Chinese, while strictly preserving all factual details such as numbers, dates, and names. The model is optimized for local execution and achieves 95% human classification by Originality.ai at its strictest setting, making it ideal for refining AI content for natural language flow.

Loading preview...

humanizer: AI-to-Human Text Rewriter

humanizer is a 12 billion parameter model developed by LucidCC13105, specifically engineered to transform AI-generated text into more natural, human-sounding prose. It supports both English and Chinese content, making it versatile for various applications. A key differentiator is its strict adherence to factual integrity, ensuring that all numbers, units, dates, names, and quotes remain unchanged during the rewriting process. The model is designed for local execution, offering privacy and offline functionality.

Key Capabilities

  • Human-like Rewriting: Rewrites AI drafts (emails, essays, reports, forum posts) to sound as if a person wrote them.
  • Fact Preservation: Guarantees that all factual elements (numbers, dates, names, quotes) are retained without alteration.
  • Multilingual Support: Handles text rewriting in both English and Chinese.
  • High Human Detection Score: Achieves 95% human classification by Originality.ai (strictest setting) on English rewrites, significantly outperforming previous versions and other tools.
  • Factual Fidelity: Maintains high factual accuracy, with 376 out of 420 English rewrites having no factual problems according to a strict LLM judge.
  • Optimized Quantization: Features advanced quantization techniques for smaller GGUF files (down to 3.9 GB for 2-bit) that maintain high agreement with the full bf16 model, enabling efficient local deployment on devices with limited memory (e.g., 8 GB).
  • Local Operation: Available as a standalone app for Mac and Windows, and compatible with llama.cpp, MLX, transformers, and vLLM for offline use.

Good for

  • Content Creators: Refining AI-generated articles, blog posts, or social media updates to enhance readability and authenticity.
  • Business Professionals: Humanizing AI-drafted emails, reports, and internal communications.
  • Students and Academics: Polishing AI-assisted essays and papers to meet natural language standards.
  • Developers: Integrating into applications where AI-generated text needs a human touch, especially for agents requiring precise fact retention.

Finetunes
1 models