ContractScrub is the first benchmark specifically for evaluating LLM performance in legal contract scrubbing. It tests long-context reasoning and consistency checking across error categories like misuse of defined terms and incorrect cross-references.
HOW THIS AFFECTS YOU
●
builderYou can use this benchmark to validate the accuracy of legal-tech LLM applications.
●
founderYou can identify gaps in frontier model reasoning to build specialized legal automation products.