OpenAI launches ChatGPT for Financial Services: GPT-6 Astra, LSEG, PitchBook, and investment banking tools

OpenAI has launched ChatGPT for Financial Services, a tailored ChatGPT Work experience built around institutional research, financial modelling and client-material production rather than a generic finance-themed chatbot.

The product combines GPT-6 Astra with built-in premium financial datasets from Daloopa, PitchBook, LSEG News and Crunchbase, while also connecting to data that financial institutions already license through provider-specific entitlement integrations and a broader connector ecosystem.

OpenAI developed the initial product direction with Morgan Stanley and Evercore, focusing first on investment banking and equity research, where reliable data access and high-quality artifact generation are recurring operational bottlenecks.

The launch targets concrete workflows including value analysis, LBO modelling, buyer screening, earnings analysis and pitchbook preparation, with the research, calculations and final deliverables increasingly handled inside the same controlled workspace.

··········

⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣

CHATGPT FOR FINANCIAL SERVICES COMBINES DATA, REASONING, AND BANKING WORKFLOWS.

The product is structured around the tasks that happen before and after the model response, including obtaining entitled data, tracing figures to their original context, moving analysis into a spreadsheet, and producing documents that follow the firm's own standards.

For investment-banking and equity-research teams, that architecture is more consequential than adding another chat interface because much of the workload consists of moving repeatedly between filings, market databases, private-company information, internal documents, Excel models and presentation software.

ChatGPT for Financial Services is designed to keep more of that sequence inside one environment: the model can retrieve information from several sources, inspect figures and supporting notes, reason over the data, compare periods, generate analysis and then convert the result into work products that can be reviewed by the team.

The Morgan Stanley and Evercore design partnership also gives the launch a narrower initial scope than the name might suggest, because OpenAI is starting with investment banking and equity research before expanding the product into additional financial-services categories.

··········

⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁤

PITCHBOOK, DALOOPA, LSEG NEWS, AND CRUNCHBASE FORM THE BUILT-IN DATA LAYER.

The native data layer is one of the most specific differences between this product and ordinary ChatGPT access because several premium datasets are indexed and hosted on OpenAI infrastructure rather than requiring the user to configure a separate MCP connection for every research request.

At launch, OpenAI names Daloopa, PitchBook, LSEG News and Crunchbase among the built-in sources, covering areas such as company financials, private-company profiles, financing activity, business news, funding histories, investors and acquisitions.

Hosting these datasets directly allows the product to return granular citations that point to specific tables and passages, so an analyst reviewing an adjusted EBITDA figure or another normalized metric can inspect the reconciliation and supporting notes instead of accepting an uncited number generated by the model.

OpenAI is also working on shared sign-in and entitlement integrations with S&P Capital IQ, LSEG, MSCI, Dow Jones Factiva and Moody's, allowing providers to recognize access rights through the user's ChatGPT sign-in where the institution already maintains a subscription.

........

Provider / layer	Access mode	Main data / function	Typical workflow
Daloopa	Built-in premium data	Financial statements and selected company metrics	Fundamental analysis and model inputs
PitchBook	Built-in premium data	Private-company profiles and financing activity	Private-market research and buyer screening
LSEG News	Built-in premium news	Financial and business news	Market context and event research
Crunchbase	Built-in premium data	Companies, funding, investors and acquisitions	Deal and private-company mapping
S&P Capital IQ, LSEG, MSCI, Factiva, Moody's	Shared sign-in / entitlement integration	Existing institutional subscriptions	Access to data already licensed by the firm
50+ connector ecosystem	Connectors / optimized MCP access	Examples include Datasite, Box, Preqin, FactSet and Intapp	Internal documents and external provider workflows

........

Data Studios mapping of the launch material identifies four named built-in data partners and five named subscription-entitlement providers before the broader ecosystem of more than 50 connectors is counted, creating at least nine specifically named provider relationships across the two most tightly integrated access layers.

That separation is operationally important because native datasets, entitled third-party subscriptions and connector-based sources do not have the same latency, contractual rights or retrieval behavior, even when they appear inside the same ChatGPT interface.

··········

GPT-6 ASTRA CONNECTS FINANCIAL REASONING TO EXCEL, DOCUMENTS, AND PITCHBOOKS.

OpenAI positions GPT-6 Astra around three capabilities required for institutional finance work: information retrieval, financial reasoning and artifact generation, with the model expected to move from source material to calculations and then into deliverables without forcing the analyst to rebuild the work manually in another application.

Astra can interpret figures, tables and supporting notes in financial documents, trace values across periods, work with annotations in public data and combine information from multiple sources before producing an analysis.

The output layer is equally central to the product because completed analysis can be turned into documents, spreadsheets, slides, interactive charts and visualizations while keeping the underlying data and sources available for review.

Administrators can publish approved Excel, Word and PowerPoint templates through a dedicated admin interface, together with firm style guides, so the system can generate valuation models, research notes and pitchbooks in formats already used inside the institution rather than returning a generic document that analysts must restyle from scratch.

For banking teams, this makes artifact quality a measurable part of model performance: a strong answer is less useful if the spreadsheet breaks established modelling conventions, the pitchbook ignores the firm's layout, or the analyst cannot trace a valuation input back to the underlying table and note.

··········

SECURITY, DATA RIGHTS, AND FRESHNESS LIMITS REMAIN PART OF THE FINANCIAL MODEL.

ChatGPT for Financial Services inherits enterprise controls from ChatGPT Enterprise, but institutions still need to map those controls onto their own information barriers, retention rules, data licenses and review procedures before using the system for material transaction or investment work.

The product supports SAML SSO, SCIM provisioning and role-based access controls; business data is not used to train OpenAI models by default, data is encrypted at rest and in transit, workspace retention can be configured, and supported logs can be exported through the OpenAI Compliance Platform into existing audit and investigation processes.

Access to skills and apps can also be managed by role, supported read and write actions can be enabled or disabled, and multiple workspaces can be used to enforce information barriers between teams or activities that should not share the same data context.

........

Control area	Built-in capability	Institution still needs to validate
Identity and permissions	SAML SSO, SCIM, role-based access	Role mapping, least privilege and approval workflows
Business data	Encryption and no model training by default	MNPI policy, retention and permitted data classes
Information barriers	Role controls and multiple workspaces	Separation of teams, deals and restricted information
Audit and investigation	Supported log export through Compliance Platform	Archiving, surveillance and internal-control integration
Source reliability	Granular citations and source context	Timestamp, delay, calculation and license checks

........

OpenAI's Financial Services terms make the remaining data-quality constraint explicit: data and model output can be inaccurate, incomplete, delayed or out of date, so users are expected to review sources, timestamps and calculations before relying on them.

The timing can vary materially by dataset; the terms specify that Nasdaq market data supplied through the service is delayed by at least 15 minutes and that Daloopa data carries a 24-hour delay, which means the presence of a source inside the interface should not be interpreted as proof of real-time coverage.

The same terms state that Financial Services is intended for information, research and analysis rather than financial or investment advice, while partner-data licenses can impose additional restrictions on downloading, redistribution, model training and the creation of substitute datasets.

··········

AVAILABILITY AND PRICING KEEP THE LAUNCH FOCUSED ON INSTITUTIONAL DEPLOYMENT.

ChatGPT for Financial Services is currently available to eligible financial institutions through OpenAI's sales and account channels, and the launch does not publish a standard list price that can be compared directly with ordinary ChatGPT subscriptions.

That procurement model is consistent with the product architecture because the economic value depends on more than model access: each deployment can involve data entitlements, existing provider contracts, workspace governance, compliance integration, firm templates and the specific workflows the institution wants to move into ChatGPT.

The important comparison for a bank or investment firm is therefore not the cost of one model response but the total cost of producing a reviewed output, including retrieval, data licensing, analyst verification, modelling, formatting and governance; a system that removes interface handoffs but increases checking time would not produce the same economic result as one that preserves traceability throughout the workflow.

The launch gives OpenAI a materially different position in enterprise finance because GPT-6 Astra is being packaged with licensed data, entitlement handling, firm templates, citations and enterprise controls rather than sold as an isolated reasoning model.

Whether that stack becomes a primary research surface will depend on data coverage, source freshness, artifact accuracy and the amount of review required before a model, memo or pitchbook can enter an institutional decision process.

··········

FOLLOW US FOR MORE.

·····

DATA STUDIOS

·····

datastudios.org