CompassRank

CompassRank is the official LLM leaderboard system of OpenCompass, developed by Shanghai AI Laboratory, that presents comprehensive rankings for comparing large language model performance across both open-source and proprietary benchmarks. It aggregates evaluation results from 70+ datasets covering dimensions such as knowledge, reasoning, coding, mathematics, and language, enabling industry-wide model assessment. As part of the OpenCompass 2.0 ecosystem alongside CompassKit and CompassHub, CompassRank supports public submissions from model developers and researchers who wish to have their models ranked. The platform is designed to provide fair, open, and reproducible benchmarks for the AI community to understand and compare LLM capabilities.