On September 10, 2026, OpenAI launched ChatGPT for Financial Services - a ChatGPT Enterprise-based industry SKU with GPT-6 Astra, firm templates, citation-backed research, and premium datasets (Daloopa, PitchBook Essentials, LSEG News, and more) indexed and hosted by OpenAI. It is not plain ChatGPT Work productivity, and OpenAI has not published a public seat price. Access is through sales for eligible financial institutions.
In short
- Finance SKU launched Sep 10, 2026; built on ChatGPT Enterprise (OpenAI launch post)
- Initial focus: investment banking and equity research (design partners Morgan Stanley and Evercore)
- Bundled selected datasets - not full vendor terminals; not on standard Enterprise
- Firm Excel / Word / PowerPoint templates plus source citations with highlights
- Workspace-wide plan; you cannot mix Enterprise and Financial Services seats
- Access via sales / eligible FIs; no public price - contact OpenAI
- High-level fork vs Claude for Financial Services and Microsoft 365 Copilot
What is ChatGPT for Financial Services vs ChatGPT Enterprise or Work?
ChatGPT for Financial Services is a separate plan for financial institutions that sits on ChatGPT Enterprise controls. Per OpenAI's Help Center overview, it provides a tailored ChatGPT Work experience inside that finance SKU - with built-in financial data and firm templates - not the kitchen-table flyer-and-deck flow of plain ChatGPT Work.
For product context on ChatGPT itself, see the ChatGPT directory page. For the everyday productivity story this SKU is not, see ChatGPT Work: finished docs without coding.
What bankers actually get in the FS plan:
- GPT-6 Astra as the model (newer models as OpenAI releases them)
- Selected premium datasets indexed and hosted by OpenAI, on by default for included sets
- Firm templates for Excel, Word, and PowerPoint that admins publish
- Citations that point to tables and passages with highlights for review
- Enterprise-grade SSO, SCIM, RBAC, retention, and Compliance Platform log export
Reuters and VentureBeat both corroborated the Sep 10 launch, the named data partners, and the design-partner framing with Morgan Stanley and Evercore.
Who inside a bank or fund actually uses this?
OpenAI's initial focus is investment banking and equity research. That shapes who gets value first.
| Role | What they actually get | Buyer note |
|---|---|---|
| IB / ER analysts & associates | Research companies, analyze earnings, compare peers, test assumptions; build editable models, notes, and pitchbooks with firm templates and citations | Primary operator seats |
| Coverage / deal teams (VP+) | Review layer using highlighted citations and underlying tables - aid, not advice | Judgment and quality control |
| Workspace / ChatGPT admins (IT) | Enable FS workspace-wide; set Financial services permissions; publish templates; control skills/apps and connector read/write; use multiple workspaces for info barriers | Enablement and policy |
| Compliance / risk / info-sec | SSO/SCIM/RBAC, retention, Compliance Platform logs, no-training-by-default, partner-data terms | Governance checklist |
| Procurement / data licensing | Bundled selected hosted sets ≠ full terminal; entitlement connectors may still apply for deeper coverage | License gap analysis |
Who it is not yet: every financial-services desk. OpenAI has stated intent to expand beyond IB/ER; treat that as roadmap, not a promise that every desk is covered on day one.
Which data sources are bundled, and how do citations work?
Included means selected coverage that OpenAI indexes and hosts. Those sets are enabled by default. For those included sets, firms do not need separate provider contracts or connector setup. Availability and coverage can change, and selected coverage may differ from each vendor's full commercial product.
| Partner / source | What you get (as published) | Caveat |
|---|---|---|
| Public company filings | SEC filings | Included |
| Quartr | Earnings-call transcripts, investor presentations, international filings | Included |
| Daloopa | Financial statements and selected metrics | 24-hour delay; Help lists 3,000 datapoints per user per month |
| Fiscal.ai | Company financials, fundamentals, ratios, operating metrics | Included |
| PitchBook Essentials | Foundational private-company profiles and recent financing | Essentials - not full PitchBook |
| Crunchbase | Private-company profiles and historical funding | Included |
| LSEG News | Reuters financial and business news | Included for US-based financial professionals (eligibility rules apply) |
| FMP / Nasdaq | U.S. equity pricing and market data | Nasdaq data delayed 15 minutes |
Citations are meant for review: figures and claims can point to specific tables or passages with highlights. Some web sources open as links. OpenAI and the Financial Services Terms are clear that this is not investment advice - verify before you use numbers in a client-facing deck.
Bundled financial data sources are not available on standard ChatGPT Enterprise. Soft rule: do not say "all of LSEG" or "full PitchBook." LSEG News is a news slice; broader LSEG sits in the entitlement / connector path.
What still needs entitlement connectors or separate licenses?
OpenAI describes more than one data path. Beyond the hosted bundle, it says it is working on shared sign-in / entitlement integrations with providers including S&P Capital IQ, LSEG (broader), MSCI, Dow Jones Factiva, and Moody's. Treat those as in progress / roadmap, not a fully shipped checklist, until your admin UI and Help Center say otherwise.
Separately, OpenAI points to optimized MCP / connectors (for example S&P Global and FactSet) and a broader connector roster that can include names like Datasite, Box, Preqin, and Intapp. Connector lists expand - attribute and recheck.
Why users get asked to connect a provider: deeper or full coverage beyond the selected hosted sets. If your desk lives in FactSet or Capital IQ all day, plan for entitlement connectors and existing licenses - do not assume the bundle replaces those terminals.
What governance controls matter for IT and compliance buyers?
The SKU inherits Enterprise-style controls and adds finance-specific admin surfaces.
Baseline controls (Help + launch):
- SAML SSO, SCIM, RBAC
- Encryption at rest and in transit
- Admin-managed retention
- Business data not used to train by default
- Log export via the OpenAI Compliance Platform
Admin path (documented): Workspace settings → Financial services → workspace default and role overrides → enable ChatGPT for Financial Services and financial data source permissions.
Firm templates: Admins publish Excel, Word, and PowerPoint templates and style guides so pitchbooks and notes match house format.
Info barriers: Multiple workspaces can separate teams. Skills and apps can be gated by role; read/write on supported apps can be controlled.
Legal layer: Financial Services Terms (published Sep 10, 2026) apply. Outputs are not financial or investment advice. Partner redistribution and training restrictions matter for compliance review.
Seat rule that trips buyers: the plan applies to the entire workspace. You cannot mix standard ChatGPT Enterprise seats and Financial Services seats in one workspace. Plan migrations and pilot scopes accordingly. If you are still weighing firm-wide Enterprise-class seats versus lighter plans before that cutover, see small business vs enterprise AI seats.
How does this compare at a high level to Claude for Financial Services and Microsoft Copilot?
This is a fork for buyers, not a scoreboard. For a broader ChatGPT vs Claude framing outside finance SKUs, see Claude vs ChatGPT.
| Path | Rough shape | What you still license |
|---|---|---|
| ChatGPT for Financial Services | Hosted selected datasets + firm templates on Enterprise controls; entitlement / MCP for deeper coverage | Deeper provider products beyond the hosted slice |
| Claude for Financial Services (Anthropic) | Connector / MCP ecosystem angle for finance analysis (Anthropic materials from 2025 onward) | Provider connections and enterprise deployment choices |
| Microsoft 365 Copilot finance | Federated connectors inside Excel / Chat / Researcher workflows | Separate provider licenses for LSEG, Moody's, Daloopa, PitchBook, FactSet, S&P Global, and similar |
No invented ranking, cost-per-task winner, or OfficeQA tournament claim belongs here. Vendor-reported evals exist in launch coverage - treat them as vendor-reported, not independent procurement rankings.
FAQ
Can I get Daloopa, PitchBook Essentials, or LSEG News on standard ChatGPT Enterprise?
No. OpenAI's Help Center states that the bundled financial data sources are not available on standard ChatGPT Enterprise. Those selected datasets ship with the Financial Services plan, where OpenAI indexes and hosts the included coverage.
Is the bundled data real-time, or are there delays?
It is not uniformly real-time. Delays vary by source. Help and Terms call out a 24-hour delay for Daloopa (plus a published datapoint limit per user per month) and a 15-minute delay for Nasdaq market data via FMP. Always check the current Help table before promising live ticks to a desk.
Can we mix ChatGPT Enterprise seats and Financial Services seats in one workspace?
No. The Financial Services plan applies workspace-wide. OpenAI documents that Standard Enterprise and Financial Services seats cannot be mixed in the same workspace. Pilots usually mean a dedicated FS workspace or a full workspace cutover plan.
Who is eligible, and is there a public price or waitlist?
Eligible financial institutions contact OpenAI Sales or their account team. Help says availability depends on organization size and type plus data-provider restrictions (for example LSEG News rules for US-based financial professionals). OpenAI has not published a public seat price or a public waitlist in the materials reviewed - do not invent either. Ask sales for current commercial terms; pricing and eligibility can change.
Does ChatGPT for Financial Services replace our FactSet or S&P Capital IQ terminal?
Not as a drop-in. The hosted bundle is selected coverage, not each vendor's full commercial product. Broader S&P Capital IQ, FactSet-class depth, and similar terminals still sit in entitlement connectors, MCP, and your existing licenses. Use the SKU as a governed research and templating layer that can sit beside those systems - not as an automatic terminal replacement.
Next step
Map seats by role (analyst → deal team → admin → compliance). Confirm which workflows fit the hosted bundle versus which still need entitlement connectors. Bring sales a checklist: workspace cutover (no seat mix), SSO/SCIM/RBAC, retention and Compliance Platform logs, template ownership, partner-data terms, and license gaps for FactSet / Capital IQ / broader LSEG.
OpenAI has not published public pricing. Confirm eligibility and commercial terms with OpenAI before you commit seats. Rates and data coverage may change.
