To pass the prestigious three-part Chartered Financial Analyst (CFA) exam, humans typically spend nearly 1,000 hours studying over several years. But new research shows that AI-powered models have now advanced to the point where they can pass even the most difficult Level III mock exams in just a matter of minutes.
In a new study conducted by researchers from New York University’s Stern School of Business and Goodfin—an AI-driven wealth management platform—23 large language models were evaluated on their ability to answer multiple-choice and essay questions from mock CFA Level III exams. The study found that frontier reasoning models such as o4-mini, Gemini 2.5 Pro, and Claude Opus successfully used “chain-of-thought prompting” to achieve passing results.
Earlier research, particularly two years ago, had shown that AI could pass Levels I and II but struggled with Level III due to its heavy reliance on essay-based questions. However, with rapid technological advancement, researchers wanted to test whether these models could now handle the “specialized, high-stakes analytical reasoning required for professional financial decision-making.” The Level III exam primarily focuses on portfolio management and wealth planning.
“I believe there’s a very real future where this technology transforms the industry,” said Anna Zhu Fee, founder and CEO of Goodfin, who contributed to the research but did not fund it.
Still, Fee emphasized that she does not believe AI will ultimately replace CFAs.
“There are elements like context and intent that are still difficult for machines to evaluate,” she said. “Humans excel at interpreting body language and subtle cues.”