fwwrsd/qwen-qwen3-vl-4b-instruct

VISIONConcurrency Cost:1Model Size:4BQuant:BF16Ctx Length:32kTool Calling:SupportedPublished:May 30, 2026License:apache-2.0Architecture:Transformer Open Weights Cold

Qwen3-VL-4B-Instruct is a 4 billion parameter vision-language model developed by Qwen, designed for comprehensive multimodal understanding and generation. It features superior text understanding, deep visual perception, extended context length, and enhanced spatial and video dynamics comprehension. This model excels in visual agent tasks, visual coding, advanced spatial perception, and multimodal reasoning, making it suitable for complex AI applications requiring robust visual and textual processing.

Loading preview...

Qwen3-VL-4B-Instruct: A Powerful Multimodal Vision-Language Model

Qwen3-VL-4B-Instruct is the latest 4 billion parameter vision-language model from Qwen, offering significant upgrades in multimodal capabilities. This model integrates superior text understanding and generation with advanced visual perception and reasoning, supporting an extended context length of 32K tokens. It introduces several key enhancements, including robust visual agent functionalities for GUI interaction, improved visual coding capabilities to generate Draw.io/HTML/CSS/JS from visual inputs, and advanced spatial perception for 2D/3D grounding.

Key Capabilities

  • Visual Agent: Operates PC/mobile GUIs, recognizes elements, understands functions, and completes tasks.
  • Visual Coding Boost: Generates code (Draw.io/HTML/CSS/JS) directly from images or videos.
  • Advanced Spatial Perception: Judges object positions, viewpoints, and occlusions, enabling stronger 2D/3D spatial reasoning.
  • Long Context & Video Understanding: Features native 256K context, expandable to 1M, for handling extensive documents and hours-long video with full recall.
  • Enhanced Multimodal Reasoning: Excels in STEM/Math tasks, providing causal analysis and logical, evidence-based answers.
  • Upgraded Visual Recognition: Broad and high-quality pretraining allows it to recognize a vast array of entities, from celebrities to landmarks.
  • Expanded OCR: Supports 32 languages and is robust against low light, blur, and tilt, with improved parsing for rare characters and long documents.
  • Seamless Text-Vision Fusion: Achieves text understanding on par with pure LLMs through unified comprehension.

Good For

  • Applications requiring complex visual and textual interaction, such as intelligent assistants or automated UI operations.
  • Code generation from visual designs or mockups.
  • Tasks demanding deep spatial reasoning and embodied AI.
  • Processing and understanding long-form video content or extensive documents.
  • Multimodal problem-solving in scientific or mathematical domains.
  • Advanced OCR needs across multiple languages and challenging conditions.

Finetunes
1 models