How much does Claude cost for a team?

What Anthropic publishes per seat

Claude for a business is sold per seat per month in two commercial tiers, Team and Enterprise. Free, Pro, and Max are consumer plans for one person. They are not a cheaper version of Team. Central billing, single sign-on, role-based access, and an admin console start at Team, which means a workforce running on personal Pro logins is a workforce you cannot see, provision, or offboard.

The prices below were on Anthropic's pricing page and support articles at the time of writing, for US customers, excluding tax. Anthropic states on those same pages that price and plans are subject to change at its discretion. Open the live pricing page and confirm the figure before it goes into a budget request, because a wrong number does more damage to your credibility than a high one.

Note the trap in the Enterprise price. Twenty dollars a seat looks identical to an annual Team Standard seat, and it buys something entirely different. On Team, the seat fee includes an allowance of usage. On Enterprise, Anthropic is explicit that the seat fee covers access only and includes no token allowance, so all consumption bills separately at API rates. Same headline number, very different shape on a budget line.

One place Anthropic's own pages do not line up. The Enterprise solutions page states a flat minimum of 20 seats. The Enterprise support article gives two minimums depending on how you buy: 20 seats self-serve, 50 seats sales-assisted. If you are sizing a sales-assisted deal, plan against 50 and confirm it with the rep, because the public page will tell you 20. Both pages are linked in the sources below.

• ·Claude Free: $0, individual account, no central administration
• ·Claude Pro: $17 per month on an annual subscription at $200 billed up front, or $20 billed monthly, still an individual account
• ·Claude Team, Standard seat: $20 per seat per month billed annually, or $25 billed monthly
• ·Claude Team, Premium seat: $100 per seat per month billed annually, or $125 billed monthly, described by Anthropic as five times the usage of a Standard seat
• ·Claude Enterprise: $20 per seat per month, annual billing only, with all usage billed separately at API rates
• ·Seat counts: Team supports 2 to 150 seats, self-serve Enterprise starts at 20, sales-assisted Enterprise starts at 50
• ·Above 150 seats you are on Enterprise whether or not you wanted the usage-based model

Seats and API usage are two different budget lines

People conflate two products that share a brand. A claude.ai seat is a subscription for a named human who logs into the Claude app. The Claude Platform API is metered consumption priced per million tokens, used by software you build or buy. If one person in the room is quoting you a per-token rate and another is quoting a per-seat rate, they are pricing different things and both can be correct. Decide which one your project is before you argue about the number.

On Team, each member gets their own usage limits that reset weekly, and one person hitting a limit does not reduce anyone else's. If an owner enables usage credits, members can keep working past the included limit and that overflow is billed at standard API rates. If usage credits are off, they stop and wait for the reset. That switch is the most consequential setting in your billing console, and the person who owns the budget should decide it deliberately rather than find out about it when someone gets cut off mid-deliverable.

On Enterprise there is no included allowance to run out of, so usage bills from the first token. Self-serve Enterprise buys usage as credits up front, drawn from a single shared pool, and Anthropic is direct that when the pool empties, usage stops for the whole organization until an owner buys more. Sales-assisted Enterprise is billed monthly in arrears, so nothing stops and the invoice grows instead. Both support spend limits at the organization and individual level, and Anthropic notes that setting a limit to unlimited does not turn billing off.

• ·Per-seat and per-token are separate lines and can both appear on the same bill
• ·Team: seat fee includes usage, overage is optional, and overage bills at standard API rates when usage credits are enabled
• ·Enterprise: seat fee includes no usage, so every token bills at API rates from day one
• ·Self-serve Enterprise credits are shared across the organization, so one heavy user drains the pool for everyone
• ·Sales-assisted Enterprise does not stop when spend climbs, it invoices in arrears
• ·Set organization and per-user spend limits before rollout, not after the first surprising invoice

Annual versus monthly, and what happens when someone leaves

Annual saves 20 percent on a Team Standard seat, $20 against $25. Anthropic's pricing page states that annual subscriptions are available for Pro and Team, and that Enterprise is annual only. The discount is real and so is the tradeoff, because mid-term flexibility runs in exactly one direction.

Adding runs smoothly. Anthropic's own billing example: an annual Team plan with 10 Standard seats is charged $2,400 up front on the start date, and upgrading five of those seats to Premium in April is prorated across the remaining 275 days and charged immediately. Removing does not work the same way. Anthropic states that removing a member gets you no immediate credit or refund. The seat becomes available to reassign, and your bill only drops if you reduce the plan's total seat allocation. On self-serve Enterprise, seats cannot be removed mid-term at all.

If you run a service business with normal turnover, that asymmetry is the thing to plan around. Buy the seats you will keep occupied, add mid-term once demand proves itself, and treat seat count as a ratchet you can only loosen at renewal. A department head who over-orders in month one is committing you for the year, and the person who finds out is you, in month seven, reconciling a headcount report against a seat report.

• ·Team Standard annual: $20 per seat per month, billed up front for the year
• ·Team monthly: $25 per seat, charged at the start of each cycle for the members on the plan at that moment
• ·Adding a seat, or upgrading Standard to Premium, is prorated and charged immediately
• ·Removing a member produces no credit or refund; reduce total seat allocation to lower the bill
• ·Self-serve Enterprise seats cannot be removed mid-term
• ·Enterprise has no monthly option

Who actually needs a paid seat on day one

Not everyone, and the instinct to buy a seat for all 200 people so nobody feels left out is the most expensive way to find out what you needed. Seats are easy to add mid-term and hard to remove mid-term. That asymmetry alone argues for starting narrow and expanding on evidence.

Pick the first group by task shape, not by title. The people who benefit first are the ones whose work is mostly reading, writing, and deciding from documents, who have a task they repeat every week, and who will tell you honestly whether it helped. A billing coordinator who works the same three claim exceptions every Tuesday is a better first seat than a director who is enthusiastic in the meeting and opens Claude twice. Anthropic's usage settings show month-to-date spend per member, which is the honest basis for deciding later who needs a Premium seat instead of guessing up front.

Then be clear-eyed about the people without a seat. On Team, a member with no seat assigned stays in your organization but cannot use Claude at all. Some of them will go get a personal Free or Pro account, which sits outside your admin console, your single sign-on, and any agreement you signed. If that is not acceptable for your data, the choice is to buy them a seat or to actually enforce the block. Both have a cost. Pretending the question will not come up is the option that costs the most.

• ·Start with a pilot of roughly 10 to 25 people doing document-heavy work, not one volunteer from each department
• ·Weight the pilot toward repeating weekly tasks, because that is where a change in hours is measurable
• ·Keep everyone on Standard seats first and buy Premium only where month-to-date usage shows a real ceiling
• ·Leave the rest as members with no seat assigned rather than paying to park them
• ·Decide in advance what the unseated majority may do, because personal consumer accounts are the default answer if you do not decide

The costs that are not the license

Multiply seats by twelve and you get a number finance will approve. It is not the cost of the program. The rest of it is internal time, it is real, and it is the difference between a rollout that holds at renewal and one that quietly dies.

Somebody has to own this. Not sponsor it, own it. That person answers the questions, maintains the shared projects and prompts, watches usage, decides who moves to a Premium seat, and follows up with the people who tried it once, got a mediocre answer, and stopped. In a company of a few hundred, that is a standing fraction of one person's week, indefinitely. If you do not name them, the honest answer is that nobody does it, and the tool becomes a subscription nobody can defend.

The gap between Claude being impressive in a demo and useful on a Tuesday is context. Your intake process, your matter or job numbering, your fee schedule, the tone you use with clients, the four things your firm never puts in writing. Getting that into shared projects and skills is specific to you, it is genuine work, and it does not stay done. It drifts every time a template, a policy, or a payer rule changes. Budget the maintenance, not just the build.

Then there is training people who will not read documentation. They will not read the wiki. They will copy whatever they watch a colleague do. What works is short, live, inside the real workflow, and repeated, and it costs staff hours plus the trainer's. The common failure is not refusal. It is someone trying the tool on the wrong task, getting a generic answer, and concluding it does not work here, which is a conclusion that spreads faster than any enablement plan.

None of this is an argument against buying it. It is an argument for putting the whole number in the request the first time. A program that quietly consumes several times its license cost in unbudgeted internal time is the one that gets cut at renewal by a finance team who were told it cost $20 a seat.

• ·An owner with explicit hours, not a line in someone's objectives
• ·Prompt, project, and skill building, specific to your firm and requiring upkeep as templates and policies change
• ·Training that is short, live, repeated, and delivered inside the real workflow, priced at staff hours
• ·IT and security work: single sign-on setup, connector decisions, and whoever answers the client security questionnaire afterward
• ·Measurement: pick two or three numbers before you start, because without them renewal becomes an argument about impressions
• ·A reserve for usage overage in the first quarter, since your early estimate of consumption will be wrong in one direction or the other

What people ask next

• We already have Microsoft Copilot. Why would we add Claude?
• How do I roll out Claude to my whole team?
• What is the difference between Claude Team and Enterprise?
• Can I control what my staff put into Claude?

Worked examples