NEW YORK CITY — OpenAI unveiled ChatGPT for Financial Services, a version of its AI product tailored for banks and investment firms, on September 10, 2026. The company developed the product in collaboration with Morgan Stanley and Evercore to integrate artificial intelligence into institutional workflows.

The specialized platform is powered by OpenAI's GPT-6 Astra model, which was launched on September 3, 2026. Access to the service requires a ChatGPT enterprise account and is limited to eligible institutions that must contact OpenAI directly for clearance.

Nick Turley, OpenAI’s vice president and head of ChatGPT, stated that finance is one of the industry verticals OpenAI has chosen to build specific products for, along with cybersecurity and software engineering. "This is the canonical product we are hoping the industry adopts," Turley said during a press briefing.

Turley described the operational demands placed on financial professionals as a driver for the tool's development. "If you study the life of an analyst or of a banker, depending on the industry, they're working 100-hour weeks," he stated. He added that the company is focused on changing how work is performed rather than simply accelerating existing processes.

Joseph Kim, Product Lead for ChatGPT for Financial Services at OpenAI, emphasized this shift in approach. "We’re trying to think of new ways of doing the work, rather than just making the existing ways faster," Kim said. The product allows users to specify whether the Astra model should approach tasks with high, medium, or low effort levels, which affects token usage and cost.

ChatGPT for Financial Services integrates data from S&P Global, Moody’s, and PitchBook, in addition to previously disclosed sources like Daloopa, Crunchbase, and LSEG News. Users can also connect existing subscriptions to datasets from Bloomberg and FactSet through the platform. The system offers approximately 50 connectors through MCP, an open-source protocol for connecting AI models to other software and data sources.

OpenAI reported that connector error rates for financial data providers improved after the company's optimization efforts. Error rates for Quartr dropped from 5.09% to 1.99%, while S&P Global saw a reduction from 6.84% to 2.66%. FactSet error rates decreased from 9.59% to 6.45%, and Daloopa improved from 7.53% to 2.57%.

The platform can generate PowerPoint presentations, Excel spreadsheets, and web-based dashboards. Company administrators can pre-load the product with branded presentation templates to maintain institutional standards. "Teams can now quickly conduct deep research across multiple sources and create detailed artifacts in one shot," OpenAI stated.

Data within ChatGPT for Financial Services is encrypted and protected via SAML SSO, SCIM provisioning, and role-based access controls. These security measures are designed to meet the compliance requirements of regulated financial institutions.

In the OfficeQA Pro benchmark, GPT-6 Astra reportedly achieved a 69.9% correctness score on U.S. Treasury Bulletins. This performance outpaced GPT-5.6 Sol, which scored 60.2%, and Claude Fable 5.1, which recorded 62.4%.

The release follows Anthropic's debut of a version of Claude for financial analysis in July 2025. That competing product included finance-specific agents for workflows such as pitchbook generation and KYC screening. Chris Churchman, a partner at Goldman Sachs, warned that the automation of tasks that help train junior bankers risks causing "cognitive atrophy" in the next generation of financiers.

Why It Matters

The launch of ChatGPT for Financial Services marks an expansion of OpenAI's enterprise offerings into a highly regulated and data-intensive sector. By collaborating with major financial institutions like Morgan Stanley and Evercore, OpenAI aims to establish its platform as a standard tool for institutional research and analysis. The integration of diverse data sources and the implementation of robust security protocols address key barriers to AI adoption in finance.

Improvements in connector error rates demonstrate the technical refinements made to ensure data accuracy, a critical requirement for financial decision-making. The competitive landscape includes other major AI providers, such as Anthropic, which entered the market with its own financial tools in 2025. As OpenAI's enterprise revenue surpasses its consumer business, the company is prioritizing specialized products that deliver measurable efficiency gains for professional users.