#financial-statements
Balance sheet, income statement, and cash-flow generation research
MultiHiertt (ACL 2022) introduces 10,440 QA pairs from real financial reports averaging 3.89 hierarchical tables each; state-of-the-art models score 38% F1 versus 87% for humans, with a 15-point penalty for cross-table questions — quantifying the retrieval gap finance AI must close.
FinanceBench evaluates 16 AI configurations against 10,231 questions from real SEC filings; shared-vector-store RAG answers correctly only 19% of the time, and even GPT-4-Turbo with the oracle passage reaches just 85% accuracy — showing that numerical reasoning, not retrieval, is the binding constraint for enterprise finance AI.
FinMaster (arXiv:2505.13533) оценява o3-mini, Claude 3.7 Sonnet и DeepSeek-V3 в 183 финансови задачи — разкривайки, че моделите постигат 96% на финансова грамотност, но падат до 3% при генериране на отчети, като многоетапните консултантски задачи губят 21 пункта точност поради разпространение на грешки.
