zenlm/zen3-vl

VISIONConcurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Feb 24, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

zenlm/zen3-vl is an 8 billion parameter vision-language model, repackaged from Alibaba Qwen's Qwen3-VL-8B-Instruct. This model supports text, image, and video modalities, making it suitable for image understanding, optical character recognition (OCR), and complex visual reasoning tasks. It is a permissively-licensed redistribution within the OSS-clean Zen model line, not trained from scratch.

Loading preview...

zen3-vl: A Vision-Language Model

zenlm/zen3-vl is an 8 billion parameter vision-language model, repackaged from Alibaba Qwen's Qwen/Qwen3-VL-8B-Instruct. This model is part of the Zen LM line, developed by Hanzo AI, and is provided as a permissively-licensed redistribution, not trained from scratch.

Key Capabilities

  • Multimodal Understanding: Processes and generates content based on text, image, and video inputs.
  • Architecture: Built on the Qwen3-VL architecture (Qwen3VLForConditionalGeneration).
  • Parameter Count: Features 8 billion dense parameters.
  • License: Distributed under the apache-2.0 license, retaining the upstream license from Alibaba Qwen.

Use Cases

  • Image Understanding: Analyze and interpret visual content.
  • Optical Character Recognition (OCR): Extract text from images.
  • Visual Reasoning: Perform complex reasoning tasks that combine visual and textual information.

This model is ideal for developers seeking a robust, open-source-friendly vision-language model for applications requiring multimodal input processing.