Julius Research releases DeepFinance financial benchmark
Understand this story
Julius Research releases the DeepFinance benchmark and leaderboard, designed to measure AI models' ability to complete real workflows in private equity and investment banking. The benchmark contains 21 tasks, each requiring an average of 27.1 hours for a human professional to complete, and the tasks were written and reviewed by PE/IB professionals from institutions including J.P. Morgan, UBS, Evercore, and Harvard Business School. In testing, each harness enables nearly all native tools (including web search), with only the ask-user tool disabled, so that each run can conclude on its own; failed attempts are rerun, and each configuration is scored on each task based on 10 successful attempts.
AI overview based on source reports · Updated 47 min ago
Coverage timeline
Follow the coverage to explore different perspectives.
- Hacker News AIDeepFinance: Can AI diligence a $100M deal?
Julius Research introduces DeepFinance, an ultra long-horizon benchmark of 21 tasks built from real, anonymized Private Equity and Investment Banking workflows that take professionals 20–30 hours each, averaging 19 files per data room and 84 rubric criteria per task. Grading covers output, workflow and judgment, and on the leaderboard GPT-6 Astra (max) leads at 57.6%, followed by Claude Opus 5.5 (54.8%), Claude Sonnet 5.5 (54.6%) and GPT-6.1 Sol (54.3%), which are not statistically distinguishable, while the bottom group scores 10.9% to 14.5%. The authors report that agents often find decision-relevant evidence without acting on it and execute calculations correctly on wrong inputs, and argue that auditability of the trajectory matters as much as final numbers.
Activity for this story
Not enough continuous observations to show a trend yet.