🌟 WritingBench: A Comprehensive Benchmark for Generative Writing
Evaluating LLMs' writing capabilities across 1,000 real-world queries, spanning 6 primary domains and 100 fine-grained subdomains.
📅 Last Updated: 2025-11-27 | 📈 Scored by: Claude-Sonnet-4-5
10 | claude-3-7-sonnet-20250219-thinking | Mistral AI | 83.87 |
10 | claude-3-7-sonnet-20250219-thinking | 83.87 | 84.46 | 83.78 | 82.21 | 85.26 | 85.64 | 82.34 |
10 | claude-3-7-sonnet-20250219-thinking | 83.87 | 91.67 | 87.09 | 86.50 | 82.00 | 89.80 | 82.17 | 80.55 | 83.67 | 82.50 | 85.75 | 82.00 | 84.00 | 86.17 | 82.40 | 83.56 | 84.25 | 79.00 | 89.20 | 86.20 | 85.50 | 88.73 | 79.50 | 85.83 | 82.91 | 88.40 | 80.50 | 86.25 | 87.82 | 75.00 | 82.17 | 83.06 | 84.15 | 87.29 | 75.14 | 84.25 | 78.40 | 82.91 | 88.67 | 87.20 | 84.00 | 84.40 | 82.62 | 81.56 | 88.91 | 82.83 | 86.33 | 82.22 | 80.00 | 72.50 | 77.33 | 81.14 | 81.33 | 83.38 | 78.36 | 79.71 | 80.33 | 74.33 | 91.71 | 88.80 | 90.18 | 89.00 | 84.50 | 84.73 | 80.33 | 86.50 | 83.50 | 87.25 | 85.60 | 76.73 | 87.64 | 86.73 | 86.22 | 87.67 | 86.00 | 83.83 | 76.67 | 68.00 | 89.71 | 88.46 | 91.25 | 88.67 | 88.86 | 84.89 | 83.75 | 84.25 | 86.00 | 81.43 | 75.00 | 79.00 | 88.46 | 82.62 | 87.29 | 86.36 | 80.62 | 83.45 | 84.13 | 77.33 | 82.50 | 76.29 | 74.00 |
10 | claude-3-7-sonnet-20250219-thinking | 83.87 | 83.09 | 83.45 | 83.86 | 85.15 | 81.91 | 75.62 |
Requirements: Style / Format / Length
Requirement_R: Includes all queries related to a specific requirement along with their entire set of criteria.
Requirement_C: Includes all queries related to a specific requirement but only the criteria pertinent to that specific requirement.
title={WritingBench: A Comprehensive Benchmark for Generative Writing},
author={Yuning Wu and Jiahao Mei and Ming Yan and Chenliang Li and Shaopeng Lai and Yuran Ren and Zijia Wang and Ji Zhang and Mengyue Wu and Qin Jin and Fei Huang},
year={2025},
url={https://arxiv.org/abs/2503.05244},
}```