TreasuryBench

TreasuryBench offers a comprehensive evaluation of personal finance applications, comparing their ability to answer real-world money questions. Key aspects include: • Grounding: Accuracy in utilizing personal financial data • Correctness: Factual precision in responses • Resolution: Effectiveness in fully addressing user queries • Prudence: Reliability and safety of financial advice provided This benchmark rigorously tests platforms like Treasury, Origin, and Monarch using identical financial scenarios and transaction histories. An independent judging mechanism assesses performance across 12 distinct categories. The methodology, scoring code, and raw results are openly available for full transparency. TreasuryBench serves as a vital resource for individuals and teams seeking to evaluate and select the most reliable financial guidance tools. It helps users understand which applications provide the most accurate, helpful, and safe answers to their personal finance questions, ensuring they can make informed decisions without risk.