The short answer
Gemini Enterprise has three separate seat ceilings, and only one of them is about your headcount. Since August 19, 2026 the Google Cloud console caps self-serve and resold accounts at 25 seats per account and invoiced accounts at 1,000. Separately, the Business edition is documented as "1-500 users", and the Frontline edition cannot be bought at all until you already hold 150 or more Standard or Plus seats. Crossing the 25 seat line does not raise your price, it changes how you buy: you place an offline order through a Google Cloud sales representative or request an allowlist exemption.
Last updated: August 2026. Seat rules and quotas read from Google Cloud documentation on August 29, 2026.
Most pricing questions about Gemini Enterprise are asked in dollars. The ones that actually derail a rollout are asked in seats, and they have nothing to do with the per-user rate. Google publishes four different numbers that constrain how many licenses you can hold and how you are allowed to acquire them, and three of the four are recent enough that the guides currently ranking for this topic do not mention them.
This is written for the team that has run a successful pilot and now wants to expand it. That is the moment these limits bite, and it is a bad moment to discover a procurement dependency.
How many seats of Gemini Enterprise can I buy without contacting sales?
Twenty five. Google documents that when you purchase or modify a seat-based subscription directly through the Google Cloud console, self-serve and resold billing accounts are limited to a maximum of 25 seats per account. Invoiced accounts get 1,000. Above either line, Google's own instruction is to contact your Google Cloud sales representative or account manager to purchase additional licenses through an offline order, or to request an allowlist exemption.
The wording that matters is what happens if you do not plan for it. Google refers to a maximum seat limit error that appears when you try to add licenses to an existing subscription. You do not get a warning at seat 20. You get a failed transaction at the point you were trying to onboard a department.
| Limit | Value | What it actually constrains |
|---|---|---|
| Self-serve or resold account, via console | 25 seats | How you buy. Past this you need an offline order or an allowlist exemption |
| Invoiced account, via console | 1,000 seats | How you buy. Large deployments move to offline orders |
| Business edition | 1 to 500 users | Which edition you are eligible for. Past 500 you move to Standard or Plus |
| Frontline edition | 150+ Standard or Plus seats required | Whether you can buy it at all. It is a companion tier, not an entry point |
What does it cost to add more seats?
Per seat, nothing changes. There is no published step-up, no volume band and no penalty rate at 26 seats. What changes is the transaction. A console purchase is immediate and self-service. An offline order is a quote, a negotiation and a signature, and it runs on your account manager's calendar rather than yours.
Model that as time, not money. If your plan says a 30 person pilot becomes 120 people in the same quarter, the 25 seat cap sits between those two states and there is no way to click past it. Teams that budget six weeks of procurement lead time are fine. Teams that assumed a card payment are the ones who slip.
Worth saying plainly: an offline order is often the better outcome. Negotiated Google Cloud pricing routinely beats list, so a company forced into a sales conversation at 60 seats may well pay less per seat than it would have paid at 25. The cost is calendar time and internal effort, not dollars.
Is Gemini Enterprise Business limited to 500 users?
Yes. Google's editions comparison lists Business as "1-500 users", and the Business edition help page describes adding seats "for a total of up to 500 seats." Unlike a console cap, this one is a product boundary: past 500 people you are on Standard or Plus, both of which Google documents as "1 or more users" with no published upper bound.
Compare that with how Microsoft handles the same shape of decision. Copilot Business stops at 300 users and crossing it forces a 67 percent per-seat jump to the enterprise add-on. Google's 500 line is softer, because moving from Business to Standard is an edition change rather than a change of base plan family, but you should still know which side of it your three year headcount plan lands on before you commit.
Does Gemini Enterprise have a seat minimum?
Not on the mainstream editions. Standard, Plus and Pay-as-you-go are all documented as "1 or more users." The only minimum Google publishes is on Frontline, and it is unusual: Frontline requires "150 or more users of Standard or Plus editions."
Read that carefully, because it is easy to misread as a cheap tier for small deskless teams. It is the opposite. Frontline is only available to organizations that have already bought 150 full seats, and it carries just 2 GiB of pooled storage with 20 Assistant queries a day on Starter or 40 on Worker. A 60 person retail operation cannot buy Frontline as its way in.
The limit that costs money is not a seat limit
Seats are the constraint people ask about. Quota is the one that generates invoices. Gemini Enterprise bills a subscription seat plus consumption, and the consumption side has a rule that makes per-seat allowances misleading: quotas are pooled. Google states that individual usage is not capped at the per-user limit and that all users in an edition draw from their edition's shared pool until it is depleted, resetting daily at midnight Pacific.
So 100 Standard seats does not buy each person 160 Assistant queries a day. It buys the edition 16,000 a day, and twenty heavy users can spend most of that before the other eighty log in. Storage works the same way: Standard includes 30 GiB pooled, Business 25 GiB, Plus 75 GiB, and the overage rate Google does publish is $5 per GiB per month, prorated by how long you store the data. An extra 100 GiB of indexed content is $500 a month regardless of how many seats you hold.
Google shipped a real answer to this on August 11, 2026: administrators on invoiced accounts can now enable overages, configure monthly spend limits and monitor feature usage. That is the right control to have. It also has a sharp edge, because a spend limit that prevents an overage prevents the work too, and an agent that stops mid-task because a cap was set conservatively is a support ticket rather than a saving. Set the ceiling with a month of real usage data behind it, and make sure the variable line lands somewhere your finance team already categorizes and reviews software spend, because a cloud invoice with two SKU families on it is easy to approve without anyone noticing which half grew.
What to check before you commit
Four questions, in the order they tend to cause trouble:
- Which account type are you on? Self-serve caps at 25 seats through the console. If your rollout is larger, start the offline order conversation before you need it, not after the error.
- Where does your headcount land against 500? That is the Business edition ceiling and it decides your edition, not just your price.
- What is your real content volume in GiB? Index a representative sample first. Pooled storage is the line most likely to produce a surprise at $5 per GiB per month.
- How uneven is your usage? A handful of power users on pooled daily quotas is the pattern that pushes a team onto a higher edition regardless of seat count.
The full edition ladder, the published and secondary price figures with sourcing on each, and the overage detail are on our Gemini Enterprise pricing page. If you are still comparing platforms rather than pricing one, Gemini Enterprise alternatives covers features and reach, Gemini Enterprise versus Glean versus Copilot puts the three big platforms side by side, and Gemini Enterprise permissions and identity sync covers the access-control work that comes with a Google Cloud deployment.
None of this makes Gemini Enterprise a bad buy. Google documents its quotas and its one published overage rate more candidly than most vendors in this category, several of whom publish nothing at all. The point is narrower: the seat number in your business case is not the number that decides whether you can execute the rollout on the timeline you promised. If you would rather not have a seat cap, an edition ceiling and a pooled quota in the same decision, the wider shortlist of enterprise search tools shows which products price on headcount alone.