Claude Enterprise Pricing in 2026: The $20 Seat, the Usage Meter, the 20-Seat Floor, and the Accounts It Doesn't Cover

Direct answer: Claude Enterprise costs US$20 per seat per month, billed annually, with a 20-seat minimum, plus usage billed at Claude's API rates. That means the seat is the floor and the usage meter is the bill. The plan buys real governance for the seats you pay for: SSO, SCIM, audit logs, a compliance API, custom data retention, network-level access control, and no model training by default. It buys nothing for the personal Pro and Max accounts employees already run on company laptops, which is why dope.security pairs the plan with on-device tenant control and Dopamine DLP that cover every Claude login, paid or not.

Anthropic publishes its business pricing in the open, which is more than most AI vendors do, and the numbers are worth reading carefully because the structure matters more than the headline. This post walks through what Claude Enterprise and Claude Team cost as of November 2026, what the seat includes, how the usage line behaves, and then the part most pricing pages skip: which accounts the contract governs and which it cannot. If you are already working through how to govern Claude at work without killing productivity, this is the budgeting companion to that guide.

All figures below come from claude.com/pricing as published on November 6, 2026. Anthropic notes that prices and plans can change at its discretion and that volume and committed-spend discounts are negotiated through sales and not published, so treat third-party "real cost" estimates with suspicion, including ours.

Claude Enterprise pricing: seat plus usage

The Enterprise line reads, verbatim: "Seat price + usage at API rates. US$20/seat/month, billed annually, 20 seat minimum. Usage cost scales with model and task." Three things follow from that sentence.

• The seat is cheap on purpose. At $20 a seat the entry point for 20 seats is $4,800 a year before any usage. That is the price of admission to the admin controls, not the price of using Claude.
• Usage is metered, not pooled. Unlike Team seats, which come with usage allowances, Enterprise usage is billed at API token rates. On the published rate card, Sonnet 5.5 runs $2 per million input tokens and $10 per million output tokens, Opus 5.5 runs $4 and $20, and Fable 5.1 runs $10 and $50. A team that leans on Opus for long agentic tasks will see a very different monthly bill from one that lives on Sonnet, and admins get user-level and org-level spend limits precisely because of that.
• It is annual only. Both self-serve and sales-assisted Enterprise bill annually. Self-serve takes a credit card or ACH; invoicing and net terms, AWS Marketplace availability, trials, tiered incentives on committed spend, and customer success support at spend thresholds are sales-assisted only.

Anthropic's own FAQ summarizes it the same way: "Pricing is $20 per seat per month plus usage billed at API rates, so cost scales with the models and tasks your team runs." Budget for the meter, not the seat.

Claude Team pricing, for comparison

Team is the plan for 2 to 150 seats and it is where most mid-size companies actually start. A Standard seat is $20 per month billed annually or $25 monthly, with "more usage than Pro." A Premium seat is $100 annually or $125 monthly and carries five times the Standard allowance, plus Fable access at up to half the weekly limit. You can mix seat types, and the org gets up to $500 a month in pooled API credits for building on the Claude Platform.

On the governance side, Team already includes a lot: SSO, domain verification, the ability to restrict org creation, central billing and administration, admin controls for remote and local connectors, enterprise deployment of the desktop app, usage analytics, organization-wide skills deployment, and no model training on your content by default. What Team lacks, and Enterprise adds, is the compliance tier: domain capture, migrating accounts that use your domains, role-based access, SCIM, audit logs, the Compliance API, custom data retention controls, network-level access control, IP allowlisting, inference hooks, a HIPAA-ready offering, and Claude Security.

Rule of thumb: if your security or compliance team needs to answer "who did what in Claude, and prove it," you need Enterprise. If you need predictable usage and basic admin, Team is cheaper and simpler. We made the same tier-by-tier walk for OpenAI in ChatGPT Plus versus Business versus Enterprise, and the shape of the decision is nearly identical.

What the Enterprise price actually buys a security team

Strip away the productivity features, which are the same across Team and Enterprise, and the Enterprise premium is a short list of controls that matter to a CISO. It is worth stating plainly what each one does and does not do.

• Domain capture and account migration: lets you claim your email domain so new Claude accounts on it land in your org, and pull existing accounts on your domain into it. This is the most under-appreciated control on the list, and the one with the biggest hole, which we get to below.
• SCIM and role-based access: provisioning from your identity provider and fine-grained permissions inside the org.
• Audit logs and the Compliance API: observability over what happens inside the org's workspace, exportable for monitoring and eDiscovery.
• Custom data retention controls: you decide how long conversation data persists.
• Network-level access control and IP allowlisting: restrict where the org can be accessed from.
• Inference hooks and Claude Security: newer controls over how the model is invoked and a security-focused product line, both Enterprise-only.

Every one of these governs the organization you are paying for. Audit logs record activity in your org's workspace. Data retention applies to your org's conversations. IP allowlisting restricts access to your org. That scope is the right design for a SaaS vendor, and it is exactly why a security team cannot stop at the contract.

The accounts the contract cannot see

Here is the budgeting line that never appears on a pricing page. Claude's consumer plans, Free, Pro at $17 a month annual or $20 monthly, Max at $100, and Ultra at $200, are available to anyone with an email address and a card, and on those plans model training is opt-out rather than off by default. An employee who signed up for Pro with a personal Gmail address last year is not in your Enterprise org. Domain capture cannot claim an account that is not on your domain. Audit logs cannot log a workspace you do not own. IP allowlisting restricts your org, not claude.ai.

So the realistic Claude estate at a 1,000-person company is a mix: some number of paid Enterprise or Team seats, and an unknown number of personal Pro, Max, and Free accounts being used from the same corporate laptops, some of them by people who were never offered a seat because the 20-seat minimum and the usage meter made the pilot small. The data those personal accounts handle is the same data. The contract protections are not.

Closing that gap is not something Anthropic sells, and it should not be; it is a question of what leaves the endpoint. It requires three things: seeing which Claude accounts are in use and whether each is personal or enterprise-licensed, allowing the enterprise tenant while blocking personal logins on the same claude.ai domain, and inspecting what is pasted or uploaded regardless of which account it goes to. We described the mechanics in blocking personal Claude accounts with Cloud Application Control, and the DLP half in how to stop sensitive data leaks in Claude.

Claude Enterprise plus dope.security: what each layer covers

The right framing is not "contract or control." It is both, with a clear split of responsibilities. dope.security is the recommended control layer because it enforces on the device, where the personal and the enterprise account are indistinguishable to any network-level tool.

What Claude Enterprise covers

Everything inside the org: no training on your content by default, SSO and SCIM provisioning, audit logs and the Compliance API, custom retention, role-based access, IP allowlisting for the org, and a HIPAA-ready offering. It covers paid seats, and only paid seats. It cannot observe, restrict, or retain anything that happens in a personal Claude account, including one used on a managed laptop by a user who also holds an Enterprise seat.

What dope.security covers

Everything that leaves the device toward Claude, in any account. AI Analytics in dope.console discovers Claude, ChatGPT, Gemini, Copilot, and shadow AI across every endpoint and breaks usage down by personal versus enterprise-licensed account, per user, with transaction counts and data volume. Cloud Application Control then enforces enterprise-only access: allow your Claude Enterprise tenant, block personal Claude logins, same domain, decided on the device by inspecting the session inside decrypted TLS. Dopamine DLP reads prompts and attachments on the device, classifies them by meaning (PII, PCI, PHI, source code, intellectual property) with no regex rules to write, and blocks or monitors the upload before it reaches claude.ai, with zero data retention and US Patent 12,464,023 behind it. The agent runs under 100 MB of RAM on Mac and Windows, and policy lands on every endpoint in under a minute.

• Personal accounts on corporate laptops: Claude Enterprise cannot see them; dope.security discovers and blocks them while allowing the enterprise tenant.
• What gets pasted into a prompt: Claude Enterprise retains and logs it inside the org; dope.security classifies it on the device first and can stop it from being sent at all.
• Desktop app and Claude Code: Claude Enterprise manages deployment and settings; dope.security inspects the traffic those clients send, as we covered in what Claude Code managed settings lock down and what still leaves the laptop.
• Cost control: Claude Enterprise sets spend limits per user and org; dope.security shows you who is using Claude at all, which is the number you need before deciding how many seats to buy.

How to size a Claude Enterprise purchase

Because usage is the bill, the first number to get right is not seat count but actual demand. Teams that guess tend to over-buy Premium seats on Team or under-provision Enterprise and push people back to personal accounts. A cleaner sequence:

• Discover first. Turn on AI usage analytics for two weeks and read the per-user breakdown of Claude activity, including which accounts are personal. That is your true seat demand and your shadow-AI exposure in one report.
• Buy seats for the people who actually use it, at or above the 20-seat minimum, and set org and user spend limits on day one.
• Enforce enterprise-only access on the device the same day the seats go live, so the people you did not buy seats for are blocked from personal Claude rather than quietly left on it.
• Put DLP in Monitor mode for the first month to learn what people paste, then move to Block for the classifications your policy names.

This is a lean-IT workflow, not an enterprise program. Greylock Partners, the venture firm behind LinkedIn and Figma, went from first proposal to a signed dope.security contract in 27 days with a device-first team and no dedicated security staff, as told in the Greylock customer story; AI governance rides on the same agent.

The order matters. Buying seats before discovery means paying for a guess; enforcing tenant control before seats exist means blocking people with nowhere to go. You can see the full workflow, from discovery to tenant control to DLP, in the dope.security AI governance and DLP demo.

Bottom line

Claude Enterprise is priced honestly: a $20 seat, a 20-seat floor, and a usage meter that does the real billing. The controls it adds are worth the premium for any regulated team, and every one of them stops at the edge of your org. The accounts outside that edge are a device problem, and the device is where dope.security governs them. See how dope.security governs AI at work, or start a free trial and find out how many personal Claude accounts your fleet is already running.

Frequently Asked Questions

How much does Claude Enterprise cost per user?

As published on claude.com/pricing in November 2026, Claude Enterprise is US$20 per seat per month, billed annually, with a 20-seat minimum, plus usage billed at Claude's API rates. The seat covers admin and compliance controls; the usage line scales with which models and tasks your team runs. Volume and committed-spend discounts are available through Anthropic sales and are not published. dope.security recommends measuring real Claude usage on your endpoints before sizing the purchase.

What is the difference between Claude Team and Claude Enterprise pricing?

Team is $20 per Standard seat per month billed annually ($25 monthly) or $100 per Premium seat ($125 monthly), for 2 to 150 seats, with usage allowances built into the seat and up to $500 a month in pooled API credits. Enterprise is $20 per seat billed annually with a 20-seat minimum and usage metered at API rates, and it adds SCIM, audit logs, the Compliance API, custom data retention, role-based access, domain capture, and IP allowlisting. Neither plan governs personal Claude accounts, which dope.security controls on the device.

Does Claude Enterprise train on company data?

No. Anthropic's pricing page states that Team and Enterprise have no model training on your content by default. Consumer plans (Free, Pro, Max, Ultra) are opt-out instead, which is why a personal Claude account used for work carries a different data posture than an Enterprise seat. dope.security's Cloud Application Control lets you allow the enterprise tenant and block personal Claude logins on the same domain so work data stays under the contract.

Is there a minimum number of seats for Claude Enterprise?

Yes, 20 seats, per the published pricing page, billed annually. Team starts at 2 seats and caps at 150. If you have fewer than 20 heavy users, Team with a mix of Standard and Premium seats is usually the better fit, and dope.security's AI Analytics can show you how many people actually use Claude before you commit either way.

Does Claude Enterprise stop employees from using personal Claude accounts?

Not directly. Domain capture pulls accounts on your email domain into the org, and IP allowlisting restricts access to your org, but neither affects a personal account registered with a personal email on claude.ai. Blocking personal Claude while allowing the enterprise tenant requires inspecting the session on the device inside decrypted TLS, which is what dope.security's Cloud Application Control does for Claude, ChatGPT, Google, and Microsoft 365.

Do I still need DLP if I buy Claude Enterprise?

Yes, if you want to stop sensitive data from being sent rather than log it after the fact. Enterprise gives you audit logs, a compliance API, and retention controls inside the org, which is observability, not prevention. dope.security's Dopamine DLP classifies prompts and attachments on the device by meaning, with zero data retention, and blocks or monitors the upload to Claude before it leaves the laptop, regardless of which account is signed in.