The short answer
Since August 24, 2026, a ChatGPT Business subscription is capped at 200 paid Standard and Premium seats in total, and OpenAI states that its support team cannot raise that limit. The cap counts every purchased seat, including unassigned ones. At the ceiling you cannot buy another seat or even approve an access request that needs one. The only route past it is ChatGPT Enterprise, which publishes no price at all. That is the expensive part. Two hundred Business seats cost a published $48,000 a year. The same 201 people on Enterprise, using the $45 to $75 per seat range that 2026 procurement reports converge on, land somewhere between $108,540 and $180,900 a year. Seat 201 is not a $20 decision.
Last updated: September 2026. Every OpenAI figure read from OpenAI's own help center and pricing pages on September 2, 2026. Enterprise per-seat ranges are third-party procurement data and are labeled as such throughout.
Seat caps are the least interesting thing on a pricing page and the most likely to break a rollout. They do not appear in the tier cards, they rarely appear in comparison posts, and they tend to be discovered by an IT admin on the day they try to add one more person. OpenAI's is unusually blunt, unusually recent, and unusually final.
This is a companion to our full breakdown of ChatGPT Enterprise pricing and ChatGPT Business seat cost. That page covers every published figure. This one covers a single question: what actually happens, operationally and financially, when a growing company runs into the 200 seat wall.
What is the ChatGPT Business seat limit?
Two hundred paid seats. OpenAI's billing documentation states that starting on August 24, 2026, the upper limit is 200 paid Standard and Premium seats total per ChatGPT Business subscription, and that its support team is unable to change your maximum seat limit. Standard and Premium seats count against the same 200, and so do seats you have purchased but not yet assigned to anybody.
That last clause is the one that catches people. The documentation is explicit that the limit includes all purchased Standard and Premium seats, including unassigned seats. If your workspace bought 30 seats ahead of a hiring plan that slipped, those 30 empty seats are still consuming ceiling. They are also still being billed, because OpenAI bills on the workspace's billable seat count rather than on whether anyone signs in.
Most vendor limits arrive with an escalation path attached: contact support, contact your account manager, request an exception. This one is written to close that door in advance. The sentence about support being unable to change the limit is not a description of policy discretion. It reads as a description of the system.
What happens when ChatGPT Business hits the seat limit?
Two things stop working and one keeps working. You cannot purchase additional seats, and you cannot approve a join request that requires another paid seat, so new hires simply queue. What you can still do is assign, reassign and swap seats you have already paid for. The workspace does not break. It freezes at its current headcount.
| Action at the 200 seat ceiling | Still possible? | What it means in practice |
|---|---|---|
| Buy seat 201 | No | The purchase path in the admin console is closed, not throttled |
| Approve a join request needing a new paid seat | No | New hires wait until capacity frees up or the contract changes |
| Ask support to raise the cap | No | OpenAI states support is unable to change the maximum seat limit |
| Reassign a leaver's seat to a joiner | Yes | Seat recycling becomes your only lever below an Enterprise contract |
| Swap a Standard seat for a Premium one | Yes | Seat types are mixable and the total, not the mix, is what is capped |
| Free up ceiling by dropping unused seats | Only at renewal | Reductions take effect at the start of the next billing cycle, with no refund |
The timing asymmetry in the last row is worth sitting with, because it is where the cap turns from an inconvenience into a real cost. Adding a seat takes effect immediately and is charged prorated on the spot. Removing one takes effect at the start of the next billing cycle. On an annual plan, OpenAI states that added seats remain committed through the end of the current annual term. So you can reach the ceiling in an afternoon and you cannot climb back down from it until renewal.
How much does it cost to move from ChatGPT Business to Enterprise?
OpenAI does not publish a ChatGPT Enterprise price, so nobody can answer this exactly. What can be stated precisely is the Business side and the shape of the jump. Two hundred Standard seats billed annually cost $48,000 a year at the published $20. Moving those same people plus one to Enterprise, at the $45 to $75 per seat range 2026 procurement reports converge on, produces an annual bill of roughly $108,540 to $180,900. That is an increase of between $60,540 and $132,900 for one additional person.
| Scenario | Seats | Rate used | Annual cost | Source of the rate |
|---|---|---|---|---|
| ChatGPT Business at the ceiling, annual billing | 200 | $20 per user per month | $48,000 | Published by OpenAI |
| ChatGPT Business at the ceiling, monthly billing | 200 | $25 per user per month | $60,000 | Published by OpenAI |
| Enterprise, low end of reported range | 201 | $45 per seat per month | $108,540 | Third-party procurement reports, not OpenAI |
| Enterprise, midpoint of reported range | 201 | $60 per seat per month | $144,720 | Third-party procurement reports, not OpenAI |
| Enterprise, high end of reported range | 201 | $75 per seat per month | $180,900 | Third-party procurement reports, not OpenAI |
| Credits for advanced features | Workspace-wide | Not published in dollars | Unknown | Set in the Order Form, on top of the seat |
We are labeling those Enterprise rows carefully on purpose, and you should treat them the same way in your own model. They are aggregated deal data from buyers and resellers. OpenAI has published no per-seat Enterprise figure, and any article presenting one as a list price is presenting somebody's negotiated outcome as a rate card. Use the range to size a budget request, then get the real number from your own quote.
The last row is the part most migration estimates leave out entirely. Enterprise workspaces buy a shared credit pool at the contract level, and credits are what pay for advanced models and features once included usage is gone. The seat is one negotiated number and the pool is a second one. Only the first behaves like a subscription.
Can I buy more than 200 ChatGPT Business seats?
No. Not by paying more, not by splitting the payment, and not by asking. OpenAI's documentation directs organizations that need a larger number of seats to ChatGPT Enterprise, and states plainly that support cannot change the maximum. The cap is a product boundary rather than a billing threshold you can buy through.
There is one workaround people reach for that deserves a warning rather than a recommendation: running two separate Business workspaces to get 400 seats. It technically sidesteps the cap and it quietly destroys the reason you bought a business plan. Two workspaces means two admin surfaces, two SSO configurations, two sets of connected apps, two audit trails, and shared context that stops at the boundary between them. You would be buying an administrative problem to avoid a sales call.
The billing mechanics also shift when you cross over. Business is self-serve on a card. Enterprise is annual and, in OpenAI's own words on the pricing page, buyers should contact sales for alternative payment options such as invoicing. That sounds like a footnote and it is actually a process change: the charge stops landing on a card statement and starts landing as an annual invoice your finance team has to route and approve, on a schedule set by procurement rather than by whoever owns the workspace.
What is ChatGPT Enterprise cost per user?
OpenAI publishes nothing. The pricing page directs Enterprise buyers to contact sales, and the plan FAQ says only that Business plans are available starting at 2 users. Every specific Enterprise per-user figure in circulation, including the widely repeated 150 seat minimum, comes from buyers, resellers and procurement databases rather than from OpenAI.
That is not a criticism, and it is worth being fair about it: quote-only enterprise pricing is the norm in this category rather than the exception. Glean publishes nothing either. Coveo publishes no tiers and no dollar figures. Google's Gemini Enterprise documentation illustrates billing with a $30 seat fee without committing to it as a rate. What makes OpenAI's version harder to plan around is the combination: an unpublished seat price plus an unpublished credit price, with a hard cap forcing the conversation on a schedule set by your own headcount.
One published discount is worth knowing about because it is large and specific. OpenAI states that nonprofits can access up to a 75 percent discount on ChatGPT Business or ChatGPT Enterprise through OpenAI for Nonprofits, with Business Standard seats at $8 per user per month billed annually. If you qualify, that changes the arithmetic above more than any negotiation will.
How to plan the crossing before you hit it
The practical work here is small and it has to happen early, because every lever gets shorter as you approach the ceiling.
Audit unassigned seats first. They count against the 200 and they are being billed. Recovering 25 empty seats buys you a quarter of hiring and costs nothing except the admin time. It is also the only capacity you can free up without waiting for a renewal.
Work out your real crossing date, not your headcount. Count everyone who will need a seat, including contractors and the teams that always get added late. If that number crosses 200 inside the term you are about to sign, start the Enterprise conversation before you renew Business, not after. Renewing Business first commits you to seats you cannot reduce until the following cycle.
Ask two questions in the first sales call. What is the seat price, and what does the credit pool cost. A quote that answers only the first is half a quote. Also ask how added seats are reconciled: OpenAI documents quarterly true-ups for standard seat-bundled Enterprise contracts, typically every three months from the workspace's own subscription start date rather than calendar quarters, and notes that adding users does not immediately increase the credit pool.
Check what you are actually buying capacity for. There is a detail in OpenAI's flexible pricing documentation that undercuts a common assumption: Enterprise and Edu plans without flexible pricing have the same per-seat usage limits as Plus for most features. An Enterprise contract does not automatically buy more usage than a consumer plan. If capacity is your reason for upgrading rather than governance, confirm which pricing model your Order Form puts you on.
The cap is the pattern, not the exception
Every vendor in this category has a threshold that reprices the deal, and almost none of them lead with it. Dropbox Dash switches plans at 100 users, where 100 seats cost $18,000 a year and 101 cost $42,420. Microsoft caps its discounted Copilot Business offer at 300 users. Google caps self-serve and resold Gemini Enterprise accounts at 25 seats and invoiced accounts at 1,000. Anthropic documents the Claude Team plan as being for teams of 2 to 150, covered in our breakdown of Claude Cowork pricing and admin controls. We collected the full set in enterprise search software for teams over 100 users.
Read them together and the lesson is consistent. In this market the sticker price tells you what a small deployment costs, and the threshold tells you what a real one costs. Comparing per-seat rates without comparing ceilings will give you a shortlist that falls apart in year two.
If you bought ChatGPT to find things, check what you are renewing
One question worth asking at the 200 seat mark is what the tool is actually being used for, because the answer changes which product you should be scaling. If most of the value is drafting, analysis and reasoning, an Enterprise contract is the right conversation. If a large share of it is people trying to find a document, a policy or a decision that already exists somewhere in the company, you may be about to pay Enterprise rates for a job the product is not really designed to do.
ChatGPT reaches company content through apps and plugins that each user authorizes individually, so the assistant sees the union of what that one person has connected and is permitted to open. That is a sound security model and a per-person coverage model. Nobody can ask a question whose answer sits in a document they have not personally connected, and no admin gets a single view of what is searchable across the company.
Unified search inverts that. InSearch indexes Drive, Slack, Notion, Confluence, Gmail, Jira and Salesforce together, returns a written answer with inline citations to the source documents, and enforces each source's own access controls at query time, so a company-wide rollout never surfaces something a person could not already open. It is $20 per user per month on Team billed annually, with no 200 seat ceiling and no repricing threshold waiting at 201. Plenty of teams run both, and that is a reasonable outcome: one tool for thinking, one for finding. The mistake is paying enterprise assistant rates and expecting company-wide search to come with it.