ChatGPT for business: Business (formerly Team), Enterprise or your own AI layer?

The short answer. ChatGPT has two business plans: ChatGPT Business, which was called ChatGPT Team until August 2025, and ChatGPT Enterprise. On both, OpenAI does not use your company data to train its models by default, and both give you single sign-on and an admin console. Business is self-serve and tops out at 200 paid seats. Enterprise adds automated user provisioning, role-based access, an audit log and data storage in Europe, among other things. What neither plan sorts out on its own: reliable answers from your own systems, work that has to run as a process, and clear rules on who may do what. For that you need a layer on your own data, alongside the licence.

This guide is for organisations of 20 to 250 people who are seriously weighing up a business ChatGPT plan. We set out what the plans offer today, based on OpenAI's own documentation, and we are equally honest about when a licence is all you need.

Why the question is on the table now

In many organisations ChatGPT is in use long before anyone has made a decision about it. Staff work from free or personal accounts, sometimes with client data in their prompts. That is shadow AI, and it is usually what really prompts the question of whether to just buy a business plan. A fair question: OpenAI says its personal plans (Free, Go, Plus and Pro) are designed for individual use, while the business plans are built for organisations.

A business plan brings that usage out of the shadows. It does not, however, solve everything behind the question. For how to channel shadow AI rather than ban it, see our piece on shadow AI in SMBs.

ChatGPT's business plans at a glance

Names first, because this is where most of the confusion starts. ChatGPT Team no longer exists under that name. OpenAI renamed Team to ChatGPT Business on 29 August 2025. It was a change of name only: existing workspaces and contracts carried on as before. If you search for ChatGPT Team today, Business is what you will end up with.

Both plans are priced per user per month. Business can be billed monthly or annually; Enterprise runs on an annual contract arranged through OpenAI's sales team. We deliberately leave out the amounts, because they change regularly. For current figures, check ChatGPT's pricing page.

ChatGPT Business (formerly Team)

• Who it is for: self-serve, from two paid seats. For subscriptions created from 24 August 2026 the ceiling is 200 paid seats; beyond that, OpenAI points you to Enterprise. You can mix Standard and Premium seats, with Premium giving each user more headroom.
• Data: OpenAI does not train on your workspace data by default and you keep ownership of your inputs and outputs, according to its enterprise privacy commitments. For GDPR purposes you can sign a data processing agreement with OpenAI.
• Administration: an admin console, central billing, domain verification and sign-in through SAML SSO.
• Integrations: apps for Google Workspace, Microsoft 365, Slack and GitHub, among others. In Business, apps are switched on by default. Admins can turn them on or off for the whole workspace, but not per role.
• Retention and access: deleted conversations are removed from OpenAI's systems within 30 days, with a few exceptions such as a legal obligation to keep them. Workspace admins can view, export and delete members' conversations.
• Not included: use of the OpenAI API. If you build your own software on the models, that is billed separately.

ChatGPT Enterprise

• Who it is for: organisations that need more seats, more control or contractual terms. Pricing on request, an annual contract, and payment by invoice is possible.
• Data: the same commitment that your data is not used for training by default. On top of that, your admins decide how long conversations are kept.
• Administration and identity: everything in Business, plus SCIM to create and remove accounts automatically from your identity provider, role-based access control, enterprise key management (EKM) and IP allowlisting.
• Oversight: through the Compliance API, admins can pull an audit log of conversations and GPTs.
• Data storage in Europe: new Enterprise customers can opt for storage in Europe (the EEA plus Switzerland) at no extra cost. Eligible customers can also keep model processing in Europe. Note that data passing through integrations with external apps or web search falls outside that guarantee.
• Integrations: a selected set of apps is enabled, new apps are off by default, and admins decide per role who can use which app.
• Support: 24/7 priority support, SLAs and bespoke legal terms.

Business or Enterprise: how to choose

For an organisation of 20 to 250 people, the decision usually comes down to five questions.

• How many people will use it? Business stops at 200 paid seats. If you are close to that, plan for growth.
• How do you manage accounts? If IT handles joiners and leavers through an identity provider, SCIM in Enterprise saves manual work and forgotten accounts. Business has SSO, but no SCIM.
• Does data demonstrably need to stay in Europe? Then Enterprise is the only one of the two with European data storage, and it is a choice you make at the outset.
• Who may use which integration? In Business an app is on or off for everyone. If only finance should reach financial data, you need role-based access, and that is an Enterprise feature.
• Do you need to be able to look back? In regulated sectors, the audit log through the Compliance API is often the deciding factor.

If none of the five applies to you, Business is a sensible, manageable starting point for most mid-sized organisations. If two or more do, a conversation with OpenAI about Enterprise is worth having.

What no ChatGPT plan solves for you

A licence determines where your conversations are stored and who can access them. It does not determine whether the answers are right for your business. That depends on four things that sit outside the subscription.

1. Answers from your own systems

ChatGPT can now search connected sources itself. With Company Knowledge, Business and Enterprise users can draw on connected apps, and it respects the permissions people already have. For sources with an off-the-shelf app, such as SharePoint, Google Drive or Slack, that is usable straight away.

It is a different story for the systems your business actually runs on: your accounting package, your ERP, your industry software, your case management or planning system. If one of those has no ready-made app, you need your own MCP server. ChatGPT can connect to custom MCP servers, but someone has to build that server, secure it, keep it running and update it whenever the source system changes. Not an objection, simply work that needs doing. For how that integration layer fits together, read MCP explained for SMBs and our page on connecting your systems.

2. Knowledge you can trust, with a source

If there are three versions of the same work instruction in your folders, an assistant will find all three. The model then picks one, or blends them, and a hallucination is never far away. Reliable answers need cleaned-up sources, version control and an answer that always points back to where it came from. For HypotheekCompany we built an internal search tool on lenders' acceptance policies that answers within 2 to 3 seconds, with the source extract and document version attached. That is not something a licence gives you; it comes from how the knowledge is organised. See also knowledge management with AI and the term RAG.

3. Processes, not one-off questions

ChatGPT is built around a person who asks a question or starts a task. Much of the time saved in a mid-sized organisation, though, lies in work that has to run the same way every day: sorting incoming requests, merging data from two systems, preparing a report. That calls for decisions no subscription makes for you. Which steps, which data, which checks, and where a person reviews the outcome (human in the loop). That is the job of AI implementation, not of a licence.

4. Rules and AI literacy

A licence is a tool, not a policy. Which data may go in, which apps are switched on, and who may read colleagues' conversations? Admins can do the latter in Business, so it is something to agree with your people up front. On top of that, Article 4 of the AI Act, as amended in July 2026, requires you to take measures to support your people's AI literacy. A subscription does not take care of that. A short AI policy, role-specific guidance and an overview of the tools you use do. More on this in our piece on Article 4 and on our AI governance page.

When to build your own AI layer alongside ChatGPT

Your own AI layer, what we call an AIOS, does not replace ChatGPT as a chat window. It is the layer underneath: your systems connected, your business context captured, and questions and tasks that run on your own data. Such a layer can work perfectly well with ChatGPT, Copilot or another model.

That layer pays off when:

• your data is spread across five or more systems and people lose a lot of time looking things up, retyping and reporting;
• your most important systems have no standard integration with ChatGPT;
• answers must be demonstrably correct, with source and version;
• you want recurring work carried out, not just questions answered;
• you want to decide for yourself which model you use and where your data is processed (sovereign AI).

A business ChatGPT plan is enough if what you are after is mainly personal productivity: writing, summarising, analysing and searching documents that already live in Microsoft 365 or Google Workspace. In that case Business or Enterprise is a perfectly good step, and you can always add a layer later. For many organisations the answer ends up being both: ChatGPT for everyone, and a layer on your own data for the processes that matter. We made the same trade-off earlier for Microsoft Copilot versus an AIOS layer.

Frequently asked questions

Is ChatGPT Team the same as ChatGPT Business?

Yes. Team has been called Business since 29 August 2025. The rename did not change features, pricing or limits, and existing contracts remained in force. Later changes to pricing and seat types are separate from the rename.

Does OpenAI train on data from ChatGPT Business or Enterprise?

Not by default. Only if you explicitly choose to share data, for example through feedback, can OpenAI use it. Data that ChatGPT retrieves through connected apps is not used for training by default either.

Can my data stay in Europe?

With ChatGPT Enterprise, new customers can opt for storage in Europe, and eligible customers can also keep model processing there. Business does not offer that option. Data that passes through external integrations or web search falls outside that guarantee.

Is ChatGPT Business enough for an organisation of 50 people?

Often, yes, if the goal is personal productivity and you have no hard requirements around data storage, role-based access or an audit log. Do agree on how it may be used, explain to your people how to work with it sensibly, and look separately at the processes that take up the most time.

Where do you start if you are not sure?

With an honest picture of where time in your organisation really goes. In an AI audit we map which questions a ChatGPT licence already answers and which call for an integration or a layer on your own data. We do not recommend what you do not need.