---
viewer: false
---

---
# 🎥 Video-Bench Dataset  
**Dataset Name**: `video-bench-model-evaluations`  
**Version**: `1.0.0`  
**Description**:  
This dataset aggregates evaluation results of **8 mainstream multimodal large models** on video understanding tasks from the [Video-Bench](https://github.com/Video-Bench/Video-Bench) benchmark. It includes **nearly 9,000 video samples** with **multi-dimensional performance metrics**.

---
Video-Bench is a benchmark tool designed to systematically leverage MLLMs across all dimensions relevant to video generation assessment in generative models. By incorporating few-shot scoring and chain-of-query techniques, Video-Bench provides a structured, scalable approach to generating video evaluation.

More details in the [paper](https://arxiv.org/pdf/2504.04907).