a benchmark designed to evaluate large language models in the legal domain.

a ground-truth-based dynamic benchmark derived from off-the-shelf benchmark mixtures, which evaluates LLMs with a highly capable model ranking while running locally and quickly.

aims to track, rank, and evaluate LLMs and chatbots as they are released.

a Swedish language understanding benchmark that evaluates natural language processing (NLP) models on various tasks such as argumentation analysis, semantic similarity, and textual entailment.

a large-scale question-answering benchmark focused on real-world financial data, integrating both tabular and textual information.

CompassRank is dedicated to exploring the most advanced language and visual models, offering a comprehensive, objective, and neutral evaluation reference for the industry and research.