Goodfin and NYU Stern Professor Srikanth Jagabathula Release Landmark Study Showing AI Models Perform at a Level Above and Beyond the CFA Level III Exam Standards

AI Wealth Platform Goodfin Unveils that AI Can Now Master the Highest Levels of Financial Certification

SAN FRANCISCO, Sept. 22, 2025 /PRNewswire/ -- Goodfin, the AI wealth platform for exclusive private market investments, has released groundbreaking research led by Shilpi Nayak, Co-Founder & CTO of Goodfin, in collaboration with NYU Stern's Professor Srikanth Jagabathula, revealing that general-purpose AI models can now pass the mock CFA Level III exam standard. This milestone in AI performance marks a major step toward making expert-level financial guidance tech-led and more broadly accessible.

Wealth management has long been low-tech and reserved for wealthier individuals, leaving sophisticated products and tailored advice in a gated, high-touch world. Goodfin's platform aims to modernize that world in the era of AI – leveraging frontier AI to innovate on, and more broadly deliver, institutional-grade financial products and digital guidance. This new study provides strong validation for that vision.

"These results are more than academic," said Anna Joo Fee, Founder and CEO, Goodfin. "They point to a very immediate future where AI will drive transformation in the wealth industry, bringing sophisticated products and strategies to a tech-native audience and enable a powerful, more convenient client experience."

The initial findings from the research paper Advanced Financial Reasoning at Scale: A Comprehensive Evaluation of Large Language Models on CFA Level III was accepted at FINLLM at the International Joint Conference on Artificial Intelligence (IJCAI) – one of the world's leading AI research conferences, ranked A* in CORE conference rank.

About the Study

This research marks the most comprehensive evaluation to date of AI performance on a financial certification exam. It benchmarked 23 leading AI models, including OpenAI's GPT-4, Google's Gemini 2.5, and Anthropic's Claude Opus 4, against the CFA Level III exam -- widely considered the most advanced credential in investment management.

While past studies showed AI could pass CFA Levels I and II, Level III remains the gold standard that most resembles the real-world practice of human experts, testing portfolio construction, ethics, and nuanced scenario analysis. Until now, even top models struggled to meet this bar.

Key Findings 

• AI now reaches professional-level reasoning: For the first time, certain frontier AI models passed the CFA Level III exam out-of-the-box, with no domain-specific training. Top performers included Claude Opus 4, Gemini 2.5 Pro, and OpenAI's reasoning models.
• Essays separate the leaders: While most models performed well on multiple-choice questions, only a few excelled at the complex essay prompts requiring analysis, synthesis, and strategic thinking.
• Prompting drives performance: Chain-of-Thought prompting—guiding the AI to explain its reasoning step-by-step—boosted essay accuracy significantly by 15 percentage points, proving that how you ask matters.
• Real-world trade-offs: Models with the highest accuracy also had the highest compute costs—up to 11x more—highlighting key considerations for scalable deployment. Gemini 2.5 Flash offered a strong performance-cost balance.
• AI isn't replacing humans—yet: A parallel user study showed that while LLMs produce consistent and precise answers, they can misinterpret intent or miss personal context. The best use case today? AI-augmented advisors providing scale with oversight.

One of the key challenges in benchmark studies is validating AI outputs. For this study, Goodfin engaged certified CFA graders to systematically evaluate and score AI responses, ensuring rigor and credibility while reducing subjective bias.

"This is a significant step forward in AI's ability to reason in highly specialized, high-stakes domains," said Professor Srikanth Jagabathula. "Passing the CFA Level III exam without specific training shows that these models are beginning to demonstrate deep generalizable intelligence. With the right design and oversight, this opens the door to a future where sophisticated financial expertise can be delivered far more broadly and affordably."

Implications for the Future of Wealth Management

Goodfin is already putting these insights into practice. Its platform leverages advanced AI and its proprietary data and infrastructure to enable investing in high-growth private market opportunities. Its system of AI agents powers a dynamic end-to-end experience, from client onboarding, to providing product information and market deep research, to tailoring guidance and taking investment actions with client consent, and to delivering post-investment management and updates. This research demonstrates the power of this technology to deliver credible, high performing solutions in financial services and its potential to redefine wealth management as more progress and usage are made.

For more information on the study, please visit www.cfabenchmark.com

About Goodfin
Goodfin is an AI wealth platform built to unlock access to private market investing and accelerate wealth for a new generation of investors. Founded in 2022, Goodfin combines frontier AI technology with domain expertise to deliver curated portfolios, intelligent analysis, and institutional-grade guidance to a community of investors who seek to own and shape the future.

Media Contact:
Tracy Rubin       
JCUTLER media group    
[email protected]      

SOURCE Goodfin