Claude Enterprise Pricing and Plans, Explained (2026)

Your CFO opens the Claude pricing page and sees one number: 20 dollars a seat. That number is real, and it is also the smallest part of the bill. Claude Enterprise pricing has two parts: a fixed fee for each seat, and a charge for what your people actually use. A seat is one person’s access. This guide explains both, compares the plans, and shows where the money really goes.

All prices below come from Anthropic’s own pages and were checked on 5 October 2026. They are in US dollars and before tax. Anthropic can change them, so follow the links before you sign anything.

Key takeaways

Claude Enterprise costs US$20 per seat per month, billed annually, with a minimum of 20 seats. Usage is billed on top, at API rates: the prices Anthropic charges for each unit of text that Claude reads or writes.^[2]
Claude Team costs US$20 per standard seat per month on annual billing, for 2 to 150 people. A usage allowance is included in the seat.^[1]
The seat fee is easy to plan. The usage is not, so it needs limits from day one.
The cost that is easiest to miss is neither of the two. It is a licence that nobody uses for real work.

What is Claude Enterprise?

Claude Enterprise is Anthropic’s plan for larger organisations. It gives every person access to Claude on web, desktop and mobile, plus Claude Code, the tool for software work.^[3] On top of that, it gives IT and Security central control: who can do what, what gets logged, how long data is kept, and how much each person may spend.

The smaller plan is called Claude Team. It is built for groups of 2 to 150 people and has fewer admin controls.^[1] Both plans come with the same promise on training: by default, Anthropic does not use your inputs or outputs from its commercial products to train its models.^[6]

Claude Enterprise pricing and plans in one table

Here are the three seat types an organisation can choose from today.

Source: Anthropic pricing pages and help centre, checked on 5 October 2026. US dollars, before tax.
	Team, standard seat	Team, premium seat	Enterprise
Price per seat, per month	$20 on annual billing, $25 monthly	$100 on annual billing, $125 monthly	$20, annual billing only
Usage	Included, up to a limit	Included, with a larger limit	Not included. Billed at API rates
Size	2 to 150 people	2 to 150 people	From 20 seats (50 through sales)
Commitment	Monthly or annual	Monthly or annual	Annual
Single sign-on	Yes	Yes	Yes
Automatic user set-up (SCIM), audit logs, spend limits	Not listed	Not listed	Yes

Two things stand out. First, the Enterprise seat costs the same as a standard Team seat. Second, an Enterprise seat includes no usage at all. Anthropic’s help centre says it plainly: the seat fee “doesn’t include any usage”, and “every token your team consumes is billed separately at standard API rates”.^[3]

Where the money goes in a Claude Enterprise rollout: the seats, the usage and the rollout work

How does Claude Enterprise pricing work?

Think of it as a phone contract. You pay a fixed fee for the line, and then you pay for what you use.

Part 1: the seat fee

The seat fee is US$20 per person per month,^[2] charged for the full year at the start.^[3] For example, 200 people at US$20 a month come to US$48,000 a year, before usage and tax. You can add seats during the year and pay for the remaining months. On a self-serve plan you cannot remove seats until the term ends.^[3]

Part 2: the usage

Every question, every document and every agent task uses tokens. A token is a small piece of text, and it is the unit AI use is measured in. On Enterprise, those tokens are billed at Anthropic’s API rates. The cost depends on which model your people choose and what they do with it. A short question costs little. An agent that reads a set of files and works through many steps uses more.

This is where the bill becomes hard to predict. Nobody can tell you your usage cost in advance, and you should be careful with anyone who tries. What you can do is measure it: run one team for four weeks with spend limits switched on, then plan from real numbers.

Two ways to buy

You can buy Enterprise online or through Anthropic’s sales team. The two routes differ.^[4]

Self-serve: from 20 seats. You pay by credit card or US bank transfer, in US dollars only. You buy usage credits in advance. When the credits run out, usage stops for everyone until you buy more.^[3]
Through sales: from 50 seats. You can pay by invoice and in other currencies. Usage is billed each month for what you used, with no built-in ceiling.

One caveat: some organisations are still on older Enterprise contracts with standard and premium seats. Those charge a flat fee per seat that includes a usage allowance.^[3] If you signed some time ago, check which model your contract uses.

Team or Enterprise: which plan fits?

The price per seat no longer decides this. Control does.

Claude Team compared with Claude Enterprise: who each plan is for and what it includes

Choose Team when one department wants to start this month, you have up to 150 users, and a predictable bill matters more than central control. Team already includes single sign-on, central billing and admin controls for connectors.^[1] Each seat carries its own usage allowance with a weekly limit.^[5]

Choose Enterprise when IT and Security need to govern the rollout. Enterprise adds the controls they ask for:^[1]

spend limits for each user and for the whole organisation;
role-based access, so each team sees only what it should;
automatic user set-up (SCIM), so people are added and removed when they join or leave;
audit logs, and an interface that lets your compliance tools monitor activity;
your own data retention rules;
network controls, such as allowing access only from your company network.

A simple rule: if you have more than 150 users, or your security team needs audit logs, you need Enterprise. If neither is true, Team is the faster start, and you can move up later.

Why does Claude Enterprise feel expensive?

The seat fee is low, so the feeling usually has one of three causes.

Usage is open-ended unless someone sets limits. Anthropic’s own advice is short: “If you want a ceiling, set spend limits.”^[3] The limits exist. Somebody has to set them.
You pay for a year at the start. The seats for the whole term are charged on day one. That is a visible sum before anyone has done any work with the tool.
The value is not visible yet. Any price feels high when nobody can say what changed. This is the real problem, and no pricing plan fixes it.

What does a rollout cost beyond the licence?

Software that is bought and not used is an old problem. Zylo, a licence management company, reports that the average organisation uses only 54% of its SaaS licences.^[9] That figure covers cloud software in general, not AI, and Zylo sells tools to fix the problem. Treat it as an indication, not a precise measure.

AI follows the same pattern. McKinsey found that nearly eight in ten companies say they use generative AI, and that just as many report no significant impact on the bottom line.^[10]

So a rollout has three costs, and only the first two are on the invoice:

The seats. Fixed and known.
The usage. Variable, and controllable with limits.
The work that makes people use it. Setting the rules once with Legal and IT. Connecting Claude to the CRM, email and documents. Training each team on its own tasks.

Without the third, the first two buy access, not results. An example from PromptPartner’s own work: at a European software company above €100 million ARR, we shipped five working tools into the systems the team already used. One of them turns a week of work on sales material into minutes. Read the case study.

How to keep the bill under control

Five steps to keep a Claude Enterprise bill under control: one team first, spend limits, the right model, a weekly look, training on real work

Start with one team. Twenty to fifty people, one department, four weeks. You learn your real usage before you commit the whole company.
Set spend limits on day one. One limit per user and one for the organisation. Raise them when the work justifies it.
Match the model to the task. A summary does not need the largest model. A contract review might. Agree on a default and say when to switch.
Look at the usage every week. Who uses it, who does not, and for what. A seat that is idle after a month needs a conversation, not a renewal.
Train people on their own work. Generic prompt training changes little. A weekly hour on real tasks changes how a team works. This is how our weekly office hours run.

Before you buy seats for everyone, it also helps to know which work is worth doing with AI at all. That is what an AI audit answers: it finds the use cases that pay off and puts a number on each one.

Data and security: what buyers ask

Is our data used for training? Not by default. Anthropic states: “By default, we will not use your inputs or outputs from our commercial products […] to train our models.”^[6] The exception: if you report feedback, for example with the thumbs up or down button, or otherwise allow it, Anthropic may use those chats.

Where is the data stored? In the United States. Anthropic’s privacy centre says “data is stored in the US” and that traffic may be routed through “select countries in the US, Europe, Asia and Australia”.^[7] Enterprise offers an option to keep processing inside the US.^[4] We did not find an option on these pages to store data only in the EU. If you need that, raise it with Anthropic before you sign.

Which certifications does Anthropic hold? SOC 2 Type I and Type II, ISO 27001:2022 and ISO/IEC 42001:2023, plus a configuration for US health data (HIPAA).^[8]

This section summarises Anthropic’s public pages. It is not legal advice, and your own contract decides what applies to you.

A certificate covers the vendor. It does not cover what your own people paste into the tool. That part is yours, and it is why the rules come before the rollout. See how we set those rules once and then go live.

Frequently asked questions

How much does Claude Enterprise cost?

US$20 per seat per month, billed annually, with a minimum of 20 seats. Usage is billed on top at API rates. Prices are before tax and were checked on 5 October 2026.

How does Claude pricing work for Enterprise?

In two parts. A fixed seat fee gives each person access. Everything they use is then billed separately, based on the model and the amount of work. Admins can set spend limits for each user and for the organisation.

Why is Claude Enterprise so expensive?

The seat itself is not. The cost that grows is usage, because it is open-ended unless an admin sets spend limits. Set them on day one and review usage every week.

Is Claude Team the same as Claude Enterprise?

No. Team is for 2 to 150 people and includes a usage allowance in each seat. Enterprise bills usage separately and adds automatic user set-up (SCIM), audit logs, role-based access, spend limits and custom data retention.

Is Anthropic’s Claude available for enterprise use?

Yes. Organisations can buy the Team plan or the Enterprise plan, online or through Anthropic’s sales team. Both include single sign-on and central administration.

Does Anthropic train its models on our company data?

Not by default. Anthropic states that it does not use inputs or outputs from its commercial products to train its models, unless you report feedback or otherwise allow it.