DavidAU/Qwen3-VL-32B-Gemini-Heretic-Uncensored-Thinking
DavidAU/Qwen3-VL-32B-Gemini-Heretic-Uncensored-Thinking is a 33.4 billion parameter vision-language model based on Qwen3-VL-32B-Thinking, fine-tuned for uncensored content generation and enhanced Gemini-style reasoning. It features a 256K context length and is designed to produce detailed, compact reasoning traces. This model excels at multimodal tasks, including image analysis and content generation across various types, without content restrictions.
Loading preview...
Model Overview
This model, DavidAU/Qwen3-VL-32B-Gemini-Heretic-Uncensored-Thinking, is a 33.4 billion parameter vision-language model derived from Qwen3-VL-32B-Thinking. It has been specifically fine-tuned to be completely uncensored for both text and image content, and to integrate a Gemini Pro Preview High thinking/reasoning style. This results in very compact reasoning traces, often 4-6 paragraphs or less, while preserving Qwen's core functionalities and metrics.
Key Capabilities
- Uncensored Content Generation: Generates any kind of content and accepts all image types without restrictions.
- Enhanced Reasoning: Features Gemini-style reasoning that significantly improves image analytics, detail extraction, and output generation. The reasoning is also temperature-stable.
- Multimodal Understanding: Inherits Qwen3-VL's advanced capabilities including visual agent operations, visual coding boost (Draw.io/HTML/CSS/JS generation), advanced spatial perception, and long context (256K native, expandable to 1M) for video understanding.
- De-censoring Metrics: Demonstrates a KL divergence of 0.0048 and a refusal rate of 3/100, indicating minimal damage to the base model while achieving significant de-censoring.
Good For
- Use cases requiring unrestricted content generation and image analysis.
- Applications benefiting from compact, precise, and high-level reasoning in a multimodal context.
- Developers seeking a powerful VL model with enhanced visual perception and agent interaction capabilities.