GPT‑5.5 scores 79/100 on finance benchmark

OpenAI said on May 15 that it had started previewing a personal-finance experience inside ChatGPT for Pro users in the United States, adding bank and brokerage account connections to its consumer chatbot. The feature lets users connect financial accounts, view a dashboard of spending and subscriptions, and ask questions based on their own balances, transactions and goals. OpenAI said the product is designed to keep users “in control of your data” while grounding answers in account-level financial context. Outside reports said the system runs primarily on GPT-5.5 and that OpenAI cited a 79 out of 100 result on a finance benchmark. ### Which ChatGPT users are getting the finance feature first? OpenAI said the rollout is a preview for ChatGPT Pro users in the U.S., not a broad release across all paid tiers. The company said users can securely connect financial accounts and then ask questions about spending, savings and financial priorities from within ChatGPT. TechCrunch reported that the account-linking layer is handled through Plaid, the financial connectivity company, and that the feature supports more than 12,000 institutions including Chase, Fidelity, Schwab, Robinhood, American Express and Capital One. (openai.com) Once linked, users see portfolio performance, spending categories, subscriptions and upcoming payments, TechCrunch reported. ### What exactly does GPT-5.5 do inside the product? TechGenyz reported on May 18 that the personal-finance experience runs primarily on “GPT 5.5 thinking,” citing OpenAI’s product materials. La Nación separately reported that GPT-5.5 is used to analyze contextual questions about a user’s money and produce personalized recommendations tied to goals such as saving or paying down debt. (techcrunch.com) OpenAI’s own product post did not prominently spell out the model name in the search snippet, but it said users can ask questions “grounded in your financial context, goals, and priorities.” That framing matches outside descriptions of a system built to answer more than generic budgeting prompts by using connected account data. ### Where does the 79 out of 100 benchmark figure come from? (techgenyz.com) TechGenyz said OpenAI claimed GPT-5.5 scored 79 out of 100 on a finance benchmark because money questions require stronger reasoning. That figure also appeared in follow-on coverage in Spanish-language outlets describing the launch. OpenAI’s public launch post available in search results focused on product capabilities, security framing and user controls rather than detailing the benchmark methodology. (openai.com) As a result, the score has been reported in secondary coverage, while the benchmark’s exact test design was not clearly laid out in the material reviewed here. ### What can users ask once accounts are linked? (techgenyz.com) OpenAI said users can ask ChatGPT questions grounded in their own financial context after connecting accounts. TechCrunch said that includes queries about spending patterns, subscriptions, portfolio performance and upcoming bills. La Nación said the tool is aimed at more contextual prompts, including questions that combine spending behavior with savings goals or loan-repayment plans. (openai.com) That moves the product beyond static dashboards and toward conversational financial guidance based on live account data, according to the outlet’s description of the feature. ### What comes next in the rollout? May 15 is the start date OpenAI gave for the preview, and the company said the feature is available to Pro users in the United States. The next concrete step is broader user testing of the preview among that paid subscriber base, with Plaid remaining the named account-connection partner in current coverage. (openai.com) (lanacion.com.ar)