---
title: PRBench
canonical_url: "https://www.modelscope.cn/datasets/ScaleAI/PRBench"
md_url: "https://www.modelscope.cn/datasets/ScaleAI/PRBench.md"
repository: ScaleAI/PRBench
last_updated: 2026-06-25
license: cc-by-4.0
storage_size: "15 MB"
downloads: 463
stars: 1
---

# PRBench

> PRBench - ScaleAI 在 ModelScope 开源的数据集。PRBench: Large-Scale Expert Rubrics for Evaluating High-Stakes Professional Reasoning

ScaleAI/PRBench 是 ModelScope 魔搭社区上的数据集，存储大小 15 MB，采用 cc-by-4.0 许可。

- **Repository**: ScaleAI/PRBench
- **License**: cc-by-4.0
- **Storage size**: 15 MB
- **Downloads**: 463
- **Stars**: 1
- **Last updated**: 2026-06-25

Source: https://www.modelscope.cn/datasets/ScaleAI/PRBench

---

# PRBench: Large-Scale Expert Rubrics for Evaluating High-Stakes Professional Reasoning

PRBench consists of:
* 1,100 expert-authored conversations across Finance and Legal domains
* 18,692 expert-curated rubric criteria (10–30 per task)
* Coverage of 114 countries, 47 U.S. jurisdictions, and 25 total professional topics.
* Hard subsets (Finance-300, Legal-250) representing the most challenging tasks

We release the Finance, Finance-Hard, Legal and Legal-Hard subsets in this release, along with our evaluation code at https://github.com/scaleapi/PRBench.

See the release for full details at: https://scale.com/research/prbench

![image](https://cdn-uploads.huggingface.co/production/uploads/617b009a7b4dce0224d6b8fb/9Gip9LfXUECzRMUy1kkyc.png)

Explore our data using our visualizer at: https://prbench-explorer.vercel.app/


![image](https://cdn-uploads.huggingface.co/production/uploads/617b009a7b4dce0224d6b8fb/LwsxCpmOrJCyI_SwD4e9I.png)
