Claude Enterprise Pricing: Plans, Seats & Costs (2026)
Oct 31, 202613 min read
Claude Enterprise is priced at $20 per seat per month billed annually plus usage at API rates, per Claude.
Treat the seat fee as a published price, and ask sales about volume discounts, much as you would when scoping automation spend with a tool like Activepieces. Budgeting from a headline seat fee is how bills go over.
Claude Enterprise pricing is the total amount an organization pays Anthropic for its company-wide Claude plan, made up of per-user seat fees, the usage limits those seats include, and any extra usage billed beyond them, all set by negotiated agreement.
Claude Enterprise pricing in one table
Every current plan, price, inclusions and limits
Read down the table below and each current Claude plan is set against four things:
- what it costs
- what it includes
- what limit Anthropic publishes
- whether I could verify the figure
Free and Enterprise carry confirmed numbers on the page, and the other plans' limits are expressed only as multiples of Pro usage.
| Plan | Monthly price (annual where listed) | Inclusions | Published usage limit | Flag |
|---|---|---|---|---|
| Free | $0 | Projects, up to 5 | Limited usage every 5 hours and every week | Verified on Anthropic's pricing page |
| Pro | Not confirmed here | Not confirmed here | Not confirmed here | Unverified |
| Max 5x | Not confirmed here | Not confirmed here | Not confirmed here | Unverified |
| Max 20x | Not confirmed here | Not confirmed here | Not confirmed here | Unverified |
| Team | Not confirmed here | Not confirmed here | Not confirmed here | Unverified |
I assembled this on 31 October 2026. The Free figures come straight from Anthropic's pricing page. I left every other cell without a number rather than guess, because a guessed price in a budget is worse than a blank one.
Before you commit, get the Enterprise price and usage terms in writing.
If you are running this arithmetic for your own team, see what the same workload costs on Activepieces.
What Enterprise adds beyond Team
Those written terms matter because, on Anthropic's plan pages, Enterprise differs from Team mainly in controls, support and contract terms rather than in the assistant itself. The table shows where the two plans part ways; use it to decide which differences you'd actually pay for.
| Team | Enterprise | |
|---|---|---|
| Seat minimum and maximum | 2 to 150 seats | 20 seat minimum |
| Admin controls | Basic workspace management | Broader controls for larger organizations |
| Security and identity | Standard sign-in and protections | Enterprise-grade identity and security options |
| Usage allowance | Plan-level limits | Usage billed at API rates |
| Support | Standard support | Higher-touch support |
| Procurement terms | Self-serve billing | Contracted terms and a quote |
Source: Anthropic's Team and Enterprise plan pages.
Past 150 seats, Team stops being an option, so a growing company gets moved to Enterprise whether or not it wanted the extras.
Where Enterprise changes day-to-day cost is in admin tooling and usage terms. Negotiated usage means your per-user allowance is a number you can ask about, not one you discover when people hit limits mid-project.
You can buy Enterprise online self-serve or through a sales conversation. That brings contract terms and a quote, but also a minimum commitment you should confirm early, because it sets your floor whether or not every seat gets used.
What Claude Enterprise costs at realistic usage
Once the commitment sets your floor, usage sets everything above it, and at realistic levels usage can outweigh the seat fee.
Replace the light, typical and heavy levels with what your own staff would actually do; they are our assumptions, not Anthropic's figures.
Claude Enterprise pricing for a 25-person team
At 25 people the seat fee is predictable and the usage layer is small in absolute terms.
Budget the seats first, then add a cushion for the two or three people who will use it all day, because they always exist.
Claude Enterprise pricing for a 100-person team
At 100 people the mix matters most. A realistic spread of light, typical and heavy users means the average hides the spread: a handful of heavy users can cost more than dozens of light ones.
Claude Enterprise pricing for a 500-person rollout
At this scale, a small error in your per-user assumption becomes a large error in the total. Being a little low on typical usage across hundreds of people costs far more than one heavy user ever could.
Plan for costs beyond seat pricing
Outside the seat price sit four things:
- usage caps
- overage billing
- seats nobody uses
- the time it takes to get people working
Usage limits and overage billing
When a team leans on Claude for long documents or coding sessions, it will hit the cap mid-task. Each seat comes with a cap on how much a person can use Claude, plus rate limits that slow heavy bursts.
Once a person passes the cap, extra usage is billed at API rates, according to Anthropic's documentation, and that's how you keep working past it.
That means your heaviest users set the bill, not the average one. Confirm the exact cap and overage terms in your contract, because I couldn't verify them here.
You pay for a floor of seats whether or not anyone logs in, because Enterprise plans typically carry a seat minimum. Every idle seat is money spent on nothing, and it hides well because the invoice looks normal.
The minimum is 20 seats, per Anthropic's pricing page.
Every idle seat is money spent on nothing, and it hides well because the invoice looks normal.
Anything your own systems or scripts do through the API is metered separately, per use; seats cover only people using Claude directly. A workflow that looks free because "we already have Claude" can quietly become its own line item, so budget it apart from seats.
Budget for Claude Enterprise rollout and training
People who aren't shown what Claude does well will either avoid it, leaving you with unused seats, or misuse it, burning through caps on poor prompts.
Plan for a real ramp-up period before you judge value. Anthropic hasn't published a standard onboarding timeline that I could verify, so ask for yours in writing.
Worth checking against a plan that does not meter every step: one credit covers a whole run on Activepieces.
How Claude Enterprise compares with ChatGPT and Gemini
Compare what each one will commit to in writing, not what its page implies, because all three vendors steer you toward a sales conversation before you see a real number.
The table shows where each vendor gives you a figure and where it hands you a contract, so you know which questions to send first.
| List price | Seat minimum | Usage limits | Admin and security | Ecosystem fit | Quote required | |
|---|---|---|---|---|---|---|
| Claude Enterprise | Self-serve or sales-assisted | 20 seat minimum | Usage billed at API rates | Single sign-on and admin controls | Teams building on Claude models | Yes |
| ChatGPT Enterprise | Sales-led | Set in your contract | Defined by contract | Single sign-on and admin controls | Teams building on OpenAI models | Yes |
| Gemini for Workspace | Bundled with Workspace plans | Follows your Workspace plan | Defined by plan | Inherits Workspace admin console | Companies on Gmail, Docs and Drive | Check for add-ons |
On Google's consumer Gemini page you'll find a free plan with a Google Account and 15 GB of storage.
Limits decide your real bill more than the seat fee does. Per Docs, ask each vendor which models a seat includes, such as Claude Sonnet 5.5, GPT-6.1 Sol or Gemini 3.8 Flash, and what happens when a heavy user runs out.
If the answer is metered overage, budget for your busiest ten percent, not your average user.
Choose Gemini if your staff already live in Google's email and documents, because you add little new to administer.
Choose Claude or ChatGPT if you want the assistant to be a separate, deliberate tool, and pick the one whose models your people prefer in a trial.
Cut Claude Enterprise spend by right-sizing seats
With limits and overage rates fixed in writing, the biggest saving left is making sure every seat belongs to someone who uses it, on a plan their work needs.
Audit usage before renewal
Do this monthly, not only in the weeks before renewal. A seat nobody opens still bills, and if you never look, it renews silently into another term.
Pull the activity report, list who has barely logged in, and ask their managers whether the seat is needed.
Then look at the opposite end: people hitting limits constantly are the ones driving extra usage charges, and you want to know who they are before the invoice tells you.
Match Claude Enterprise or Team plans to roles
Not everyone needs the same tier. Heavy users who need Enterprise controls, such as admin features and security review, justify Enterprise seats.
Occasional users, like a manager who drafts a few emails a week, may be fine on Team seats, if your contract and the vendor's terms allow mixing. Ask about it directly.
Ask about annual terms too: committing for a year usually earns a better rate, but only after the audit, because you'd be locking in your mistakes. Commit to the headcount you can defend, not the one you hope for.
Move repetitive Claude jobs to API automation
Use a chat seat for a person asking varied, one-off questions. It's a poor fit for the same task run hundreds of times, like summarizing every support ticket.
Those jobs belong in an automation platform that calls Claude's API and writes results straight to your helpdesk, billed by use instead of occupying a seat's limits.
The second lane never needs a person, so it never needs a seat.
Your Monday plan for budgeting Claude Enterprise
With the machine work out of the way, what's left to budget is people, and five steps take you from the pricing page to a number you can defend, so you stop treating people costs as a guess.
- Pull the current pricing page and record the date. Prices and limits change, and a dated copy shows you later whether a bill moved because of your usage or because of their rate card.
- Count likely users by role: daily heavy users, occasional users, and people who will only try it once. A single headcount hides the fact that these groups consume very differently.
- Request a quote, using your role counts rather than a round company-wide number.
- Run a measured pilot for one month, logging real usage per person.
- Set a renewal review date while the contract is fresh, so the first bill gets checked against the pilot before the next commitment.
Prepare before requesting a Claude Enterprise quote
Do steps one and two first, because that homework protects you. A salesperson who hears "about the whole company" will quote for the whole company.
Questions to put to Anthropic sales
Ask what happens when a user hits a limit: does work stop, or does billing continue at an extra rate? Ask whether extra usage is capped, and who gets alerted before the cap.
Ask how seats can be reassigned if someone leaves or changes roles. Get each answer in writing, because the pilot only measures your usage and says nothing about what the contract allows.
What Activepieces does about this
The part of this bill that seat pricing handles worst is repetitive machine work, and that is where Activepieces fits.
It is an automation platform that can call Claude through the API from a workflow, so a job like summarizing every support ticket runs on metered usage and never occupies a person's seat or eats into their limits.
The seat stays with the people who ask varied, one-off questions.
The pricing model suits that split. A run counts as 1 credit however many steps it has, and unlimited flows come on every plan, including the free one.
So adding a workflow doesn't add a line item, and the number you budget is how often the work runs.
On Plus and up you can attach your own AI provider key, which puts the model cost on your provider bill, where you can see it and set your own spending alerts.
You then price that usage by the same pilot-and-measure method this article recommends for seats.
You can check the claims before you commit. The credit structure is published on the Activepieces pricing page, and the free plan lets you build a ticket-summary flow and count its runs against your pilot.
Alan and Funding Societies run production automation on Activepieces, so this isn't a toy for scripts that nobody depends on.
What it doesn't change is the Claude contract. Seat minimums, caps and overage terms still come from Anthropic sales, in writing.
Moving the repetitive jobs elsewhere shrinks the number of seats you need to defend and keeps your heaviest automated usage out of the extra usage billing that heavy seat users trigger.
Frequently asked questions
Is there a free trial of Claude Enterprise?
Don't plan around one. Enterprise is sold through a sales conversation, so any trial or pilot is something you negotiate, not something you switch on yourself.
Ask for the terms in writing, including what happens to your usage and your data when the pilot ends, because a pilot that quietly rolls into a paid contract is a bill you didn't budget for.
Does Claude Enterprise train on our data?
By default, Anthropic's commercial terms say it doesn't train its models on customer content. Don't take that from a summary, though, including this one.
Read the data-handling terms in your actual contract, and get your legal team to confirm how long your content is retained and what exceptions apply. If the contract and the marketing page disagree, the contract is what binds anyone.
Can we switch from Team to Enterprise mid-contract?
Often, yes, but whether you get credit for the unused part of your Team plan is a negotiation, not a rule. Ask before you sign anything.
If you don't, you may pay for both plans for a stretch, and the overlap is money spent on seats nobody is using.
Also ask whether your existing users and their conversation history carry over, or whether people start from scratch.
Is Claude Enterprise billed monthly or annually?
Because Enterprise contracts are typically negotiated, the billing period is part of the deal and not a fixed menu choice.
Annual commitments usually come with the better per-seat rate, but they also lock you in if your headcount shrinks.
Ask what happens to seats you add midyear and how extra usage is invoiced, since that charge can arrive on a different schedule from the seat fee.
Running the numbers
See what the same workload costs here.
Free forever plan, and every paid plan self-hosts at no extra cost.
See pricing Talk to sales
