OpenAI released its latest ChatGPT model today, and fintech F2 is an early adopter.
GPT-5.5 is the latest LLM model launched by OpenAI, providing improved coding, data analysis, research and operating software agents, the company announced today. The latest model uses fewer tokens to write more complex codes on its Codex platform.
“We’ve been testing it closely,” Don Muir, chief executive at F2, said in his LinkedIn post today. “The takeaway is simple: The performance jump is real.”
F2, founded in 2025, is an AI platform for private market investors, according to the company.
Muir wrote in his LinkedIn post that GPT-5.5, when integrated with F2’s agentic system for buyside workflows, enhances:
- Higher-confidence reasoning across complex diligence tasks;
- Synthesis across messy, multimodal data rooms; and
- Accuracy in financial and investment analysis.
A comparison of different AI models for software development accuracy (Courtesy/OpenAI)
Other fintechs took to LinkedIn to share their use of the new technology.
“We’ve had early access to the model and have been encouraged by its performance across our audit and advisory evals,” Chris Szymansky, chief technology officer at agentic AI platform for audit and advisory Fieldguide, posted on LinkedIn.
He also cited better token efficiency with GPT-5.5.
AI model arms race
With upgraded agentic AI capabilities, GPT-5.5 can aid FIs in better data analysis and software development, Jeff McMillan, founder of AI consultancy firm McMillanAI and former head of firmwide AI of Morgan Stanley, told FinAi News. “I have had a chance to play with it and it’s very, very good.”
While GPT-5.5 has bested all other models in the market as of now, it’s only a matter of time before another LLM provider releases an upgraded version of their model which might beat GPT-5.5, McMillan said.
“You will only get to see rough parity and marginal differences between different AI models,” he said. “But all providers are locked into an arms race to beat each other.”
For financial services companies, it’s not so much about which model is selected, it’s about “tying it tightly” to the tech stack, McMillan said.
“The way you integrate these models is more important than the model itself,” McMillan said.
Register here for the FinAi Lending Summit, set for Oct. 7-8 in Las Vegas.
