Benchmark RadarAI BENCHMARK PROFILE
FinProBench
FinProBench evaluates financial AI agents on professional tasks using rubrics derived from practitioner deliverables. It includes 1,723 curated deliverables across 57 occupations and an evaluation set of 20 tasks.
- Released
- 2026-08-04
- Readiness
- Paper only
- Primary field
- Finance & Economics
Why it matters
The benchmark addresses the need for evaluating financial AI agents against professional standards, offering an approach to derive rubrics from expert deliverables, which can reduce effort in rubric construction.
Motivation
Evaluating financial AI agents requires criteria aligned with real professional work.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.