prithivMLmods/Qwen3-VL-4B-Instruct-abliterated-v1

VISIONConcurrency Cost:1Model Size:4BQuant:BF16Ctx Length:32kTool Calling:SupportedPublished:Oct 15, 2025License:apache-2.0Architecture:Transformer0.0K Open Weights Gated Cold

prithivMLmods/Qwen3-VL-4B-Instruct-abliterated-v1 is an abliterated variant of the Qwen3-VL-4B-Instruct model, developed by prithivMLmods, designed for uncensored visual captioning and reasoning. This 4 billion parameter vision-language model excels at generating detailed, descriptive outputs for a wide range of visual and multimodal contexts, including complex or sensitive content. It supports diverse aspect ratios and resolutions, making it suitable for high-fidelity descriptions across various image types.

Loading preview...

Overview

prithivMLmods/Qwen3-VL-4B-Instruct-abliterated-v1 is an abliterated (v1.0) variant of the Qwen3-VL-4B-Instruct model, specifically tailored for uncensored reasoning and captioning. This model is designed to produce detailed and descriptive captions and reasoning outputs across a broad spectrum of visual and multimodal content, including complex, sensitive, or nuanced material. It maintains consistent accuracy across diverse aspect ratios and resolutions, leveraging the multimodal reasoning and instruction-following capabilities of its Qwen3-VL-4B foundation.

Key Capabilities

  • Abliterated / Uncensored Captioning: Fine-tuned to bypass conventional content filters while preserving factual, descriptive, and reasoning-rich outputs.
  • High-Fidelity Descriptions: Generates comprehensive captions and reasoning for general, artistic, technical, abstract, or low-context images.
  • Robust Across Aspect Ratios: Supports wide, tall, square, and irregular image dimensions with consistent accuracy.
  • Variational Detail Control: Produces outputs ranging from high-level summaries to fine-grained, intricate descriptions and reasoning.
  • Multilingual Output Capability: Primarily English, with adaptability for multilingual prompts via prompt engineering.

Good For

  • Generating detailed, uncensored captions and reasoning for general-purpose or artistic datasets.
  • Research in content moderation, red-teaming, and generative safety evaluation.
  • Enabling descriptive captioning and reasoning for visual datasets typically excluded from mainstream models.
  • Creative applications such such as storytelling, art generation, or multimodal reasoning tasks.
  • Captioning and reasoning for non-standard aspect ratios and stylized visual content.