ESGenius

An expert benchmark for evaluating LLM knowledge of ESG and sustainability standards.

Source-grounded questions, reproducible evaluation code, and question-level model analysis for sustainability reporting, climate, governance, and standards-driven reasoning.

Open interactive heatmap Get dataset

Publication

EMNLP 2025 Main Conference Oral

Nominated for Resource and Theme Paper Awards (Top 1%).

Paper title ESGenius: Benchmarking LLMs on Environmental, Social, and Governance (ESG) and Sustainability Knowledge

ACL Anthology Download PDF

BibTeX

@inproceedings{he-etal-2025-esgenius,
  title = "{ESG}enius: Benchmarking {LLM}s on Environmental, Social, and Governance ({ESG}) and Sustainability Knowledge",
  author = "He, Chaoyue and Zhou, Xin and Wu, Yi and Yu, Xinjia and Zhang, Yan and Zhang, Lei and Wang, Di and Lyu, Shengfei and Xu, Hong and Xiaoqiao, Wang and Liu, Wei and Miao, Chunyan",
  editor = "Christodoulopoulos, Christos and Chakraborty, Tanmoy and Rose, Carolyn and Peng, Violet",
  booktitle = "Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing",
  month = nov,
  year = "2025",
  address = "Suzhou, China",
  publisher = "Association for Computational Linguistics",
  url = "https://aclanthology.org/2025.emnlp-main.739/",
  doi = "10.18653/v1/2025.emnlp-main.739",
  pages = "14612--14653",
  ISBN = "979-8-89176-332-6"
}

1,136 questions

50 models evaluated

7 framework families

A-D + Z answer protocol

Dataset

Expert ESG knowledge, packaged for repeatable model evaluation.

The benchmark covers sustainability reporting, climate disclosure, biodiversity, energy, governance, and standards-driven ESG reasoning across IPCC, GRI, SASB, ISO, IFRS/ISSB, TCFD, and CDP sources.

Plain CSV Plain JSON Reference CSV

query_idStable question identifier

queryQuestion stem

A-DAnswer options

ZNot sure option

ref_docSource document in reference file

source_textSupporting excerpt in reference file

Results

Model performance at a glance, with question-level drill-down.

The homepage loads quickly, while the full Plotly heatmap remains available as a dedicated report for deep inspection.

Ranking bar chart for evaluated ESGenius models — Ranking view for evaluated models on the balanced ESGenius question set.

Interactive Heatmap

Inspect every model-question outcome without leaving the benchmark story.

The full report covers 50 evaluated models across 1,136 ESGenius questions, sorted by model rank and question difficulty for fast error-pattern analysis.

Correct Wrong Invalid Not sure Missing

Open full Plotly heatmap

50 models ranked top to bottom Hardest questions Easiest questions

Resources

Everything needed to inspect and reproduce the benchmark.

Read EMNLP 2025 paper Download ACL PDF DOI record View GitHub repository Evaluation guide Dataset documentation Apache 2.0 license