CEdit-Bench

**CEdit-Bench** is a comprehensive evaluation suite, first proposed in the **LongCat-Image technical report**, and developed by integrating and extending existing image editing benchmarks. We further curate new data to enhance task diversity, yielding a robust dataset of 1,464 bilingual (Chinese–English) editing pairs across 15 fine-grained task categories, providing a more holistic and rigorous standard for evaluating image editing models.

<div align="center">
<img src="assets/data_distribution.jpg" width="75%" alt="data distribution." />
</div>

An example entry is shown below:

```
{
"key": "730fc5ea-7c14-45c9-a2a8-8d802b49a610",
"input_image": bytes,
"instruction_en": "Adjust the background to a garden",
"instruction_cn": "将背景调整为花园",
"task_type": "Background Change"
}
```

**CEdit-Bench**是一款综合性评测套件，最早提出于**LongCat-Image技术报告**，通过整合并拓展现有图像编辑基准数据集开发而成。我们进一步精选新增数据以丰富任务多样性，最终构建得到覆盖15个细粒度任务类别、共计1464组中英双语编辑样本的高质量数据集，可为图像编辑模型的评测提供更为全面且严谨的标准。

<div align="center">
<img src="assets/data_distribution.jpg" width="75%" alt="数据分布" />
</div>

下方展示了一条样本示例：

{
"key": "730fc5ea-7c14-45c9-a2a8-8d802b49a610",
"input_image": "字节流（bytes）",
"instruction_en": "Adjust the background to a garden",
"instruction_cn": "将背景调整为花园",
"task_type": "背景更改"
}