Benchmark RadarFinVerBench
FinVerBench evaluates financial statement verification in LLMs, using SEC 10-K XBRL filings from 43 S&P 500 companies with a four-category error taxonomy (arithmetic, cross-statement linkage, year-over-year, magnitude perturbations).
- Released
- 2026-05-28
- Readiness
- Paper only
- Primary field
- Finance & Economics
Why it matters
Financial statement verification requires calibrated judgment under incomplete observability and realistic numerical rendering, not just arithmetic detection. This benchmark provides a reusable diagnostic subset to assess model performance and validity.
Motivation
We introduce FinVerBench, a benchmark and validity study for financial statement verification: determining whether a set of corporate financial statements is numerically consistent from the information shown to the model.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.