Short answer: ChatGPT for Financial Services, announced by OpenAI on September 10, 2026, is a ChatGPT Work experience designed for eligible financial institutions. It combines GPT-6 Astra, included financial datasets, connections to sources already licensed by the institution, and enterprise security and governance controls. The key is to separate those layers: some data is ready to use; some depends on an existing subscription or entitlement; and several details—such as pricing, broad geographic availability, and access for every ChatGPT user—were not promised at launch.
What ChatGPT for Financial Services is
According to OpenAI's official announcement, the product is a specialized ChatGPT Work experience for research, financial modeling, and client-material creation. Its initial development was shaped by a design partnership with Morgan Stanley and Evercore, with an early focus on investment banking and equity research.
The product includes GPT-6 Astra natively and, according to OpenAI, is expected to make newer models available as they are released. OpenAI also highlights multi-source research, tracing figures across periods, interpreting notes and tables, creating interactive charts, and generating documents, spreadsheets, and presentations.
Matrix: what is included, what requires a license, and what was not promised
| Layer | Launch status | What it means in practice |
|---|---|---|
| GPT-6 Astra and analysis tools | Included in the product | The model and research, analysis, and artifact-generation workflows are part of the offering for eligible institutions. |
| Premium data from partners named by OpenAI | Included in selected datasets | OpenAI cites Daloopa, PitchBook, LSEG News, and Crunchbase as examples of data that can be used without negotiating a separate contract or setting up a dedicated connector for those included datasets. |
| Data from providers the institution already subscribes to | Depends on an existing entitlement/license | OpenAI says it is working with S&P Capital IQ, LSEG, MSCI, Dow Jones Factiva, and Moody's on shared sign-in and entitlement integrations. That does not make paid subscriptions universally included content. |
| Financial connectors and MCPs | Integration available subject to source and permission | The cited ecosystem includes more than 50 connectors, including Datasite, Box, Preqin, FactSet, and Intapp. A connector's existence does not replace the provider permissions, contracts, or credentials required. |
| Real-time, unlimited data or broad redistribution rights | Not promised | Conditions vary by source. Some data is delayed, and there are restrictions on use, copying, export, storage, or redistribution. |
| Access for every ChatGPT user | Not promised | The launch states availability for eligible financial institutions through sales or an account team. |
Which data is included natively
The announcement describes built-in datasets from providers such as Daloopa, PitchBook, LSEG News, and Crunchbase. The stated coverage includes earnings transcripts, financial statements, company fundamentals, private-company information, and other inputs used in financial research. OpenAI says this data is indexed and hosted on its infrastructure to improve retrieval, latency, and granular citations, so users can trace figures and claims back to source tables and passages.
The Financial Services Terms also list partners such as Fiscal AI, Quartr, Nasdaq, and Financial Modeling Prep (FMP), alongside providers already mentioned. That list must be read with source-specific conditions: coverage and timing vary, Nasdaq market data listed in the terms is delayed by at least 15 minutes, and Daloopa data has a 24-hour delay.
There are also specific eligibility restrictions. The terms state, for example, that Reuters Ready News is available only to professional users employed by eligible financial-firm entities domiciled or incorporated in the United States. So “included” does not necessarily mean “available to every person, in every country, for every purpose.”
When an existing license still matters
Another layer is access based on rights the institution already holds. OpenAI says it is working with S&P Capital IQ, LSEG, MSCI, Dow Jones Factiva, and Moody's on shared sign-in and entitlement integrations. The goal is for the provider to recognize a user through their ChatGPT sign-in and automatically enable data that person is already entitled to access.
This is materially different from datasets included with the product. If a bank or asset manager already pays for an external database, the integration may reduce authentication and connection friction, but it does not eliminate the underlying contract. The same principle applies to the connector ecosystem: technical connectivity and the legal/commercial right to use data are separate matters.
What changes for banks and asset managers
The proposed gain is to bring intelligence, data, and deliverable production into one environment. OpenAI names use cases such as value analysis, LBO modeling, buyer screening, earnings analysis, and pitchbook preparation. GPT-6 Astra is presented as able to navigate figures, tables, and notes in financial documents, perform analysis, and synthesize results into documents, spreadsheets, and slides.
To standardize output, administrators can publish Excel, Word, and PowerPoint templates through a dedicated admin page. Teams can then move from analysis to a valuation model, research note, or pitchbook that follows their firm's own templates and style guides.
This is also why the offering is different from simply opening ChatGPT and asking a question about a company: the product combines financial sources, access rights, enterprise templates, administrative controls, and traceability of supporting evidence.
Who can use it — and what not to confuse with the Finances feature
At launch, OpenAI says ChatGPT for Financial Services is available to eligible financial institutions. Interested organizations are directed to sales or their account team, and the announcement does not publish a general price list, a worldwide rollout schedule, or complete eligibility criteria.
It should not be confused with the personal Finances in ChatGPT feature, which lets eligible U.S. users connect financial accounts through Plaid. ChatGPT for Financial Services is an institutional offering for professional financial work; Finances is designed around a user's connected personal financial context.
Governance and security: what comes from the enterprise foundation
ChatGPT for Financial Services builds on ChatGPT Enterprise controls. OpenAI cites SAML SSO, SCIM provisioning, and role-based access controls. It also says an organization's business data is not used to train models by default, is encrypted at rest and in transit, and can be governed by configurable workspace retention.
For compliance, supported workspace logs can be exported through the OpenAI Compliance Platform into audit and investigation workflows. Institutions can also manage access to skills and apps by role, enable or disable supported read and write actions, and create multiple workspaces to enforce information barriers.
Those organizational controls do not override data-provider rules. The terms make clear that Partner Data remains subject to each partner's license and restrictions, including limits on copying, downloading, storage, model training, creation of derived data, and redistribution.
What was not promised — and why it matters
- There is no promise that all data is real time. The contract itself includes delayed sources and tells users to check timestamps and delay labels.
- There is no guarantee that every response is correct, complete, or current. The terms call for reviewing sources, dates, and calculations and independently verifying prices before trading.
- There is no broad license to redistribute third-party data. An export or sharing feature does not expand rights granted by a data partner.
- There is no universal consumer access. Announced availability is for eligible financial institutions.
- There is no confirmation that every entitlement integration is fully live on day one. The launch says OpenAI is “working with” a number of providers.
- It is not financial advice. OpenAI classifies Financial Services and their outputs as informational tools, not recommendations to buy, sell, or hold assets.
Checklist for an institution evaluating adoption
- Confirm with OpenAI whether the institution is eligible and which commercial terms apply.
- Map which datasets are actually included in the workspace and which depend on existing licenses.
- Validate coverage, delay, and usage restrictions for each relevant provider.
- Define which connectors, apps, and skills each role may use and which read or write actions will be enabled.
- Configure SSO, SCIM, roles, retention, logs, and information barriers before putting sensitive data into production workflows.
- Test real workflows—such as earnings analysis, valuation, or pitchbook creation—measuring accuracy, source traceability, and the amount of human review required.
- Review Partner Data rules before copying, exporting, storing, or redistributing licensed content.
In summary
The central change in ChatGPT for Financial Services is not simply putting a new model in front of analysts. The offering packages GPT-6 Astra, built-in financial data, access to sources already licensed by the firm, connectors, and enterprise controls into one experience. For banks and asset managers, the potential value is reducing the work required to gather data, analyze evidence, and turn analysis into deliverables; the practical limits come from each dataset's rights, institutional eligibility, governance requirements, and the continuing need for human verification.
