---
title: OpenCaption-4B-VL-SFT
emoji: 📝
colorFrom: indigo
colorTo: pink
sdk: gradio
sdk_version: 6.24.0
app_file: app.py
short_description: Dense fine-grained image captioning with Qwen3-VL-4B
python_version: "3.12"
startup_duration_timeout: 30m
license: apache-2.0
---

# OpenCaption-4B-VL-SFT-v1.0

Dense, fine-grained image captioning with **OpenCaption-4B-VL-SFT-v1.0** — a vision-language model fine-tuned from Qwen3-VL-4B-Instruct for rich, structured image descriptions.

Upload an image and the model produces a detailed, structured caption with thematic sections covering subjects, backgrounds, lighting, atmosphere, and fine visual details.

## Model

- **Base**: [Qwen/Qwen3-VL-4B-Instruct](https://huggingface.co/Qwen/Qwen3-VL-4B-Instruct)
- **Fine-tune**: [prithivMLmods/OpenCaption-4B-VL-SFT-v1.0](https://huggingface.co/prithivMLmods/OpenCaption-4B-VL-SFT-v1.0)
- **License**: Apache-2.0