Cursor Enterprise pricing: what a seat costs above $40

What Cursor actually publishes

Start with the numbers that exist in Cursor's own pricing documentation, because a surprising number of comparison articles quote figures that do not. Every self-serve tier carries a dollar amount, and the upper ones are also described as multiples of agent limits.

The individual side of the ladder is covered in more depth in the Cursor pricing guide, including the two separate usage pools and what drains each one. This page is about what changes when the invoice stops being personal.

What Enterprise adds over Teams

Cursor presents Enterprise as everything on Teams plus a list. The list is worth reading as three separate purchases, because teams usually need one of the three and pay for all of them.

• The procurement bundle is invoice and PO billing, SCIM, and account management. If finance will not accept a card and IT will not accept manual deprovisioning, this is the only tier that answers, and no amount of arithmetic changes that.
• The control bundle is repository, model, and MCP access controls plus auto-run, browser, and network controls. This is what a security review asks for when it asks what an agent is allowed to reach.
• The economics bundle is pooled usage and the AI code tracking API. Pooled usage is the one that can pay for itself, and it is the subject of the next two sections.

The surcharge that only exists above individual plans

On Teams and Enterprise plans, third-party model requests carry a Cursor Token Rate of $0.25 per million tokens on top of the model’s own price. It applies when a developer pins a third-party model and when Auto Balance or Auto Intelligence routes to one. Auto Cost and Cursor’s own first-party models are exempt. On individual plans it does not apply at all.

This is a small number that behaves like a large one, because it is charged against every token a team’s agent traffic touches, and agent traffic is context-heavy by construction. A single agent run on a large repository can read hundreds of thousands of tokens before it writes a line.

The seat math at 10, 50, and 200 people

Teams Standard at $40 is exactly twice Pro at $20. That is the whole comparison for most teams, and it is worth stating without decoration before the feature lists start doing persuasion.

The premium is real and it is not a mistake. What you buy with it is one invoice instead of fifty expense reports, SSO instead of personal logins, team-wide privacy mode instead of a policy nobody can verify, and analytics instead of asking people how much they used. Those are worth money. They are just worth a specific amount of money, and the table above is what you are comparing it against.

1. Count the seats that will actually be active. Assigned seats bill whether or not anyone opens the editor. This is the single largest source of waste in every seat-based AI tool, and the arithmetic is the same one covered in the seat management guide.
2. Separate the seat decision from the usage decision. The $40 buys administration. It does not buy more model usage than Pro, and Teams Premium at 5x agent limits is a different purchase with no published price.
3. Price the surcharge before you sign. Measure a fortnight of third-party token volume on individual plans first, then multiply by $0.25 per million and add it to the seat total.
4. Decide whether pooling changes the answer. On a mixed team it usually does, which is why it sits on Enterprise rather than Teams.

Why pooled usage is the deciding Enterprise feature

Every engineering team has a usage distribution, not a usage average. Two or three people run agents all day, a middle group runs them a few times a week, and a tail barely opens the tool. When each seat carries its own allowance, the heavy users exhaust theirs and pay on demand while the tail leaves most of its allowance unspent, and the organization pays for both halves of that mistake.

Pooled usage collapses the distribution into one number. The tail’s unspent allowance offsets the heavy users’ overage, so the organization pays for what it consumed rather than for the shape of its consumption. On a team where the top decile drives most of the agent traffic, that difference can exceed the seat premium on its own.

Run your own distribution before the call. The AI team cost calculator takes a headcount and a usage mix and returns the per-seat and pooled totals side by side, which is a faster way to answer this than a spreadsheet built during a sales cycle.

Where Continuum’s org model differs

Continuum is not an editor and does not compete with Cursor’s tab completion, which remains the best in the category. It is a workbench and an org control plane over the agent CLIs a team already runs, and the commercial shape is deliberately different in three ways worth stating plainly.

If you are weighing whether to keep Cursor and add spend control around it, running Continuum alongside Cursor covers that setup. If you are weighing a move, the migration guide is honest about what you would give up. The side-by-side is on the Cursor comparison page.

Questions people ask

Keep reading

Continue through AI spend and governance with Cursor for teams: what $40 a seat actually buys.