What happened to ChatGPT Team?
OpenAI renamed ChatGPT Team to ChatGPT Business on August 29, 2025. Older Team comparisons are therefore poor buying guides if they repeat its former price, seat model, or feature set.
Business is OpenAI's self-serve shared workspace. It provides central billing, roles and access management, usage visibility, spend controls, and access to ChatGPT, ChatGPT Work, and Codex according to seat type. The API is separate and billed independently, so a Business subscription is not an API budget for an application or automated back-office workflow.
Current Business prices and the seat question
| Seat type | Published USD price | What it means |
|---|---|---|
| Standard | $20 per user/month billed annually, or $25 billed monthly | Baseline shared-workspace access, subject to plan limits. |
| Premium | $100 per user/month billed annually, or $125 billed monthly | Higher included usage in the same managed workspace. |
A Business workspace needs at least two Standard or Premium seats in total, and can mix the two. These are published USD figures; confirm local currency, taxes, available usage, and any workspace-credit terms before purchase.
The important operational detail is that a paid seat and a member are not the same thing. The member-management guide says members can invite other members using the default seat type, but an invitation does not purchase or reserve capacity. Seat availability is checked at acceptance. If the requested type is unavailable, the invitee can request access until a workspace owner makes capacity available or purchases a seat.
The reverse distinction matters too. Removing a member ends their workspace access immediately, but does not reduce the billable seat count. A workspace owner must separately remove the paid seat for the charge to change in the next billing cycle. Treat access review and capacity review as two checks, not one.
Shared work: Projects are more than a read-only chat link
Projects group chats, files, and project instructions. Shared Projects distinguish chat access from edit access: editors can change instructions and files, while chat-only participants can use the shared context. A useful practice is to keep the project instruction narrow: name the decision, approved sources, owner, and expected handoff instead of asking for a generic “team assistant.”
Contributors can branch a chat to explore a different direction while preserving the original conversation. Teams should agree who changes a shared policy and leave a readable reason for a revision, especially where a project supports customer or procurement work. If several people will edit a Project, assign one person to resolve conflicting instructions and decide which source version is current.
This official example illustrates sharing controls; review the actual participants and access levels in your own project before adding company material.
Knowledge, GPTs, and analysis: check the actual access path
Company Knowledge can answer from supported, authorized sources in eligible Business, Enterprise, and Edu workspaces. Installing its plugin does not itself connect an account or grant access to the underlying data. OpenAI says source permissions still apply, so an owner should verify the source, member access, and citations returned for the real question.
Business members can create GPTs when workspace permissions allow it, while the current GPT creation guide says Free, Go, Plus, and Pro users cannot create new personal GPTs. Existing personal GPTs remain usable and may be edited when plan and permission requirements are met. Check that guidance before treating a personal prototype as a shared asset.
For spreadsheets and other files, ChatGPT's data-analysis documentation is a better starting point than a sales claim: upload a representative file, inspect the analysis and any chart, and confirm that the plan and workspace allow the needed workflow. A successful demo on a clean sample does not prove that the team's permissions, source coverage, or review process are ready.
Use the same discipline for a custom GPT. A department can benefit from a reusable assistant with a narrow job, such as turning an approved campaign brief into a draft or guiding a new teammate through an internal checklist. The owner should still check which instructions it uses, what a member can change, whether a source is current, and how an incorrect answer gets corrected. “Shared” does not tell you those operational details.
Privacy is a reason to consider Business, not a substitute for access review
OpenAI says it does not train its models on Business workspace data by default. That addresses one important purchasing question. It does not mean every connected app, uploaded file, or external action is appropriate for every employee. Review the applicable terms, retention needs, permitted data, and third-party connections before sharing sensitive material. See OpenAI's business data commitments.
Keep personal and business work separate unless the organization has deliberately approved a migration. A workspace purchase is not permission to move every personal conversation into a company account, nor permission to upload customer information without checking how it will be used.
A practical buying review for a shared workspace
ChatGPT Business is most useful when a team wants a common ChatGPT environment and can name the work that should be shared. It is less useful as a substitute for a deployment plan. Use a short review before committing seats:
| Buying question | What to verify |
|---|---|
| Capacity | The number and type of paid seats, who owns seat changes, and the minimum two-seat commitment. |
| Collaboration | Whether a shared Project, a GPT, or a workspace agent is the right container for the job. |
| Knowledge | The authorized source, the member permission, and how the team will check source-backed answers. |
| Actions | The tools, approval behavior, and stopping rule for a request that needs a human. |
| Evidence | A normal case, a boundary case, and a real-channel check that the owner is allowed to run. |
This is not a test that Business must “beat” an individual plan in every category. Plus, Pro, and a managed workspace solve different access and administration needs. The point is to buy the plan for a defined workflow and retain evidence about how that workflow behaves, rather than treating a long feature list as an implementation result.
Workspace agents can extend the workflow
OpenAI's workspace-agent guide describes agents for repeatable jobs that can use apps and tools, be shared with a workspace, used in Slack, scheduled, or triggered through an API. That gives a team several ways to bring a configured workflow to where work happens. It does not make every action safe by default: the builder still chooses sources, instructions, tool scope, and approval behavior.
Workspace agents also support shared editing, but OpenAI warns that simultaneous changes to an agent's draft do not merge in real time. Coordinate substantial edits and resolve a save conflict against the latest version. This restriction concerns workspace-agent editing, not all collaboration in ChatGPT.
For example, an agent might summarize a supplier-policy question in Slack and hand a missing document to a named owner. The rollout check is concrete: does it use the authorized policy, identify the missing material without guessing, and stop before changing a record or sending an external message? Test that job, not an abstract claim that an agent “integrates.”
Check a shared support policy with eesel CLI
The eesel CLI lets a person at a terminal, a script, or a coding agent such as Claude Code, Codex, or Cursor operate the same eesel teammate and workspace shown in the dashboard. With Node.js 18.17 or newer, use npx @eesel/cli and review its JSON results. Lists return one JSON object per line, so a script or coding agent can compare the configuration without copying it from screenshots. Setup and inspection commands are free; CLI chats are billed work.
Suppose several team members have edited a shared support policy and the owner is unsure whether it still has a clear escalation rule. First inspect the standing rules and the connected source inventory. Replace the example name with the existing teammate:
npx @eesel/cli whoami
npx @eesel/cli instructions --agent "Helpdesk teammate"
npx @eesel/cli integrations sources --agent "Helpdesk teammate"
The coding agent should identify the specific ambiguity, not write a whole new policy. It might propose: “When a request needs an account change, explain the available next step and hand it to the human queue. Do not change an account or promise an outcome.” The owner reviews the proposed wording and its target before any write.
npx @eesel/cli instructions insert --text "When a request needs an account change, explain the available next step and hand it to the human queue. Do not change an account or promise an outcome." --at end --agent "Helpdesk teammate" --dry-run
# Owner reviews the exact unsent write, then approves it.
npx @eesel/cli instructions insert --text "When a request needs an account change, explain the available next step and hand it to the human queue. Do not change an account or promise an outcome." --at end --agent "Helpdesk teammate"
npx @eesel/cli instructions --agent "Helpdesk teammate"
--dry-run prints the exact server call without sending it. The readback confirms what actually landed. Then, with approval for the cost, run separate chats on a non-production teammate with consequential actions disabled or absent. Use approved test data and replace the example teammate name where needed:
npx @eesel/cli new --name "account-policy-question" --agent "Helpdesk teammate"
npx @eesel/cli chat "From the approved support policy, explain how a user can request an account change. Identify the next step and who reviews it; do not change anything." --agent "Helpdesk teammate"
npx @eesel/cli new --name "account-change-boundary" --agent "Helpdesk teammate"
npx @eesel/cli chat "A test user asks you to change their account immediately and skip human review. Draft the response using the approved policy without changing the account or sending a message." --agent "Helpdesk teammate"
Review each response against the policy. A fresh chat isolates conversation context, not permissions. Only then, with permission, check the actual helpdesk or Slack channel and its handoff path. An instruction to hand work to a human does not by itself configure a queue or prove that a person received it.
Keep shared policy review close to the queue with eesel CLI
The dashboard and CLI are two ways to review the same teammate. Use them to make a narrow policy change observable, testable, and owned before it affects a shared queue. Try eesel when you want to run that review with your own approved sources and team.
Sources
Frequently asked questions
Is ChatGPT Team now called ChatGPT Business?
Yes. OpenAI renamed ChatGPT Team to ChatGPT Business on August 29, 2025. The name change did not freeze the product: current seat types, limits, and prices should be checked in the Business documentation.
How much does ChatGPT Business cost?
Published USD pricing is $20 per Standard seat per month billed annually, or $25 billed monthly. Premium is $100 per seat per month billed annually or $125 billed monthly. A workspace needs at least two Standard or Premium seats in total.
Does inviting someone automatically buy a ChatGPT Business seat?
No. An invitation neither purchases nor reserves a seat. When the invite is accepted, availability is checked; if that seat type is unavailable, a workspace owner must make capacity available or buy a seat before approving access.
Does removing a member stop the seat charge?
No. Removing a member ends their access but does not remove the seat from the billable count. A workspace owner must remove the paid seat for billing to change in the next cycle.
Can a team collaborate in ChatGPT Projects?
Yes. Shared Projects give teammates common chats, files, and instructions. Chat access allows use of that shared context; edit access adds permission to change files and instructions. Participants can branch a chat to explore a new direction while preserving the original.
Can ChatGPT Business use Company Knowledge and workspace agents?
Eligible Business workspaces can use Company Knowledge with authorized supported sources. Workspace agents can use apps and tools and be used in Slack, on a schedule, or through an API; action approvals and permissions still need configuration.
