OpenAI builds a ChatGPT for Wall Street analysts
Built on GPT-6 Astra, the enterprise tool pulls in Bloomberg, FactSet and LSEG data through about 50 connectors and exports to Excel and PowerPoint.
In short
OpenAI has released ChatGPT for Financial Services, an enterprise build of GPT-6 Astra that drafts financial models, research and client decks from licensed market data.
At a glance
- Product: ChatGPT for Financial Services, built on GPT-6 Astra and sold only with an enterprise subscription.
- About 50 connectors, built on the Model Context Protocol, attach outside data and software.
- Data providers named across the two reports: Bloomberg, FactSet, LSEG, PitchBook, Crunchbase, Daloopa, S&P Global.
- Access is limited to institutions that apply directly to OpenAI; Morgan Stanley and Evercore helped develop it.
- Anthropic has sold Claude for Financial Analysis into the same market since May 2025.
OpenAI is selling a version of ChatGPT built for deal teams. ChatGPT for Financial Services runs on GPT-6 Astra and is aimed at investment bankers and equity analysts whose days go into models, research and client decks. It ships with an enterprise subscription only, and institutions have to apply before they are switched on.
What separates this from ordinary ChatGPT is less the model than what the model is allowed to read. Banks can attach their own records or licensed third-party sets, and SiliconAngle reports roughly 50 connectors built on the Model Context Protocol. Answers carry citations back to the data behind them, and the output lands as slide decks, spreadsheets or browser dashboards.
Each request can be dialed to high, medium or low effort. That setting trades response time and token spend against how hard the model works on the answer.
Both reports list FactSet, LSEG, PitchBook, Crunchbase and Daloopa. SiliconAngle adds Bloomberg; Handelsblatt adds S&P Global. Neither explains which feeds a given bank actually receives at the start.
Handelsblatt names Morgan Stanley and Evercore as the U.S. banks involved in development. OpenAI vice president Nick Turley calls it the product he hopes the industry standardizes on, and draws a line between work that demos well and work an analyst can actually hand to a client. Anthropic got to the same buyers earlier, launching Claude for Financial Analysis in May 2025.
Goldman Sachs partner Chris Churchman has warned that pushing this work to software risks “cognitive atrophy” among junior bankers. They learn the trade by grinding through exactly the research tasks now being automated. The objection is about how the next generation is trained, not about whether the answers are correct.
Neither report gives a price, a general availability date, or a count of approved institutions. Nothing in either text describes how the service meets European supervisory requirements; Handelsblatt frames it only as an effort to serve heavily regulated, data-sensitive industries. We are flagging those gaps rather than filling them.
FAQ
What is ChatGPT for Financial Services?
An enterprise edition of ChatGPT running on GPT-6 Astra for investment banks and equity analysts, building financial models, research summaries and client materials from connected market data with citations.
How do banks get access to ChatGPT for Financial Services?
For now only through an enterprise subscription plus a direct application to OpenAI, which decides which institutions are eligible; Handelsblatt reports Morgan Stanley and Evercore worked on the product.
How much does ChatGPT for Financial Services cost?
No price has been published. Neither SiliconAngle nor Handelsblatt names a figure; the only stated requirement is an enterprise subscription.
