Claude Enterprise is the plan for organizations with 20 seats or more: Anthropic's pricing page lists 20 dollars per seat per month, billed annually, plus usage billed by consumption at API rates, so the total emerges only from what you use; a tailored quote is only available through sales (as of October 2026). Claude Team starts at 2 seats and costs 20 dollars per seat per month billed annually, or 25 dollars billed monthly, each with a fixed usage allowance. On privacy, both plans sit on the same contractual level: no training on your content by default, and a data processing addendum built into the commercial terms. Enterprise adds administration features Team doesn't have, and requires the 20-seat minimum in return.
I run my business on an AI workforce built on Claude and pay for several Claude accounts. Business data sits on Claude Team with two seats, alongside Max for my own daily work, and I'm very happy with that combination; the arithmetic is in Claude Team: Cost and When It Pays Off. I have not bought Enterprise; what I write about it comes from Anthropic's pricing page and help center, not from my own use. Everything else about Claude is collected on my Claude page.
Short answer: Team from 2 seats, Enterprise from 20
In most cases the choice between Team and Enterprise is settled by headcount alone. Enterprise requires at least 20 seats according to Anthropic's help center, Team runs from 2 to 150. Below 20, Enterprise cannot be bought; above 150, Team is too small.
Sources for the prices are Anthropic's pricing page and my post Claude Pricing and Plans Explained; for the seat minimum, the Anthropic help article on Enterprise seats.
Claude Enterprise cost: seat plus usage
Enterprise pricing works differently from every other Claude plan. On Free, Pro, Max and Team you pay a fixed amount and get a usage allowance, expressed as a multiple of Pro. On Enterprise you pay 20 dollars per seat per month, billed annually, as a base fee and then every request by consumption, at API prices. The pricing page puts it as "Usage cost scales with model and task": what a seat costs depends on the model and how long the tasks run. At API rates, as of 26 October 2026, that means for example: Claude Sonnet 5.5 costs 2 dollars per million input tokens and 10 dollars per million output tokens, Claude Opus 5.5 4 and 20 dollars, Claude Haiku 5.5 from 0.10 and 0.50 dollars, and Claude Fable 5.1 10 and 50 dollars (developer platform pricing page). What separates the models is in my overview of the Claude models.
A seat includes, according to Anthropic's billing help article, the web app, the desktop app, mobile, Claude Code and Cowork. Usage is paid one of two ways: buy Enterprise yourself online, possible since 12 February 2026 and paid by card or ACH bank transfer in US dollars only, and you purchase credits up front; buy through sales, which Anthropic intends from 50 seats, and usage is invoiced monthly. Owners can set spend limits per organization, per group and per person; once a limit is reached, usage stops until the next billing period or until an owner raises it.
A worked example with list prices, as of October 2026: 20 Enterprise seats cost 4,800 dollars a year in base fees, with usage on top. 20 Team Standard seats on an annual contract also cost 4,800 dollars, but there the usage allowance is included, and since October 2026 there are 400 dollars of API credits a month on top (20 seats times 20 dollars). So the difference isn't the seat price; it's whether you want to pay for usage by consumption and whether you need the admin features. Anthropic publishes no guideline for total usage per seat, meaning chat, Cowork and Claude Code together. For Claude Code alone, its docs put the average across enterprise deployments at around 13 dollars per developer per active day (code.claude.com, as of 26 October 2026). I can't give you a figure of my own, because I don't run Enterprise myself.
New since 23 October 2026: chat and Cowork unified, switchover on 16 November 2026
Since 23 October 2026, the unified Claude that combines chat and Cowork has been generally available for Enterprise too. Every conversation starts as a chat. When a task needs code or files, Claude continues in the same conversation in an isolated cloud environment that Anthropic operates, with a virtual machine of its own per conversation. Outbound network access there is off by default for Enterprise.
For owners, one date matters most: in existing organizations the new experience starts out off and is switched on automatically from 16 November 2026. Under Organization settings > Capabilities you can turn it on early or deliberately keep it off, and a choice once made stays in place. It requires chat and cloud code execution with file creation to be on; otherwise nothing switches on. Organizations with zero data retention, the HIPAA configuration or customer-managed encryption keys don't get the new experience for now and keep working as before. Anthropic recommends starting with a pilot group through a dedicated role, watching for two to four weeks, and then expanding in waves.
Because tasks in the cloud can use more tokens than a quick chat, Anthropic names four levers for cost:
- Efficient mode: When an owner turns it on, org-wide or per role, the new experience uses only Sonnet and Haiku models, and new sessions start on Claude Sonnet 5.5 at low effort. Other products such as Claude Code are not affected.
- Model entitlements per role: You decide which role may use which models.
- Spend limits: per organization, per group and per person.
- Pooled group budgets: A group draws from one shared monthly amount.
Billing stays usage-based. According to Anthropic, Opus 5.5 costs 40 percent less than Opus 5 on typical workloads, and Sonnet 5.5 up to 30 percent less than Sonnet 5 for most work (claude.com, support.claude.com, as of 26 October 2026). My advice, explicitly without Enterprise experience of my own: make a deliberate decision before 16 November 2026 instead of letting the switchover simply happen, and set spend limits before the first long cloud tasks run.
What Team costs and what's included
Team comes in two tiers, per Anthropic's pricing page and help center as of October 2026: Standard at 25 dollars per seat per month or 20 dollars on an annual contract with 1.25 times the Pro usage, and Premium at 125 dollars or 100 dollars annually with 6.25 times. Both tiers can be mixed in the same workspace, and Claude Code has been included in Standard seats since 16 January 2026. Since October 2026, Team also includes a monthly credit for the Claude API: 20 dollars per Standard seat and 100 dollars per Premium seat, capped at 500 dollars for the whole team. Enterprise does not get this credit (support.claude.com, as of 26 October 2026). It also covers the Agent SDK and Managed Agents; what Anthropic offers under the name agents is in my post on Claude Agents. The worked example for a three-person consultancy and when Premium is needed are covered in Claude Team: Cost and When It Pays Off.
For the decision against Enterprise, what matters most about Team is what's already in it: SSO, role-based admin permissions, central billing and the same contractual level on privacy. If you buy Team, you're not giving up a privacy commitment, you're giving up administrative depth.
Claude Enterprise privacy and GDPR: what Anthropic commits to
This is not legal advice. I'm reporting what Anthropic writes in its own documents, as of September 2026, and flagging what isn't there. Whether that's enough for your case is something to settle with your data protection officer or lawyer.
First, training: Anthropic's Enterprise page states verbatim "Your prompts, data, and results are not used to train our models by default". The same commitment applies to Team. On the consumer plans it hangs on a setting inside each account, which nobody can enforce centrally. That is exactly why customer data belongs in a commercial product; the toggle, the retention periods and who is responsible are in Claude Privacy and GDPR: What Applies.
Second, data processing: Anthropic's Data Processing Addendum has been in effect since 24 February 2025, incorporates the EU Standard Contractual Clauses (modules 2 and 3) and the UK Addendum by reference, and names Ireland as the jurisdiction. For Team and Enterprise it is part of the commercial terms; you sign nothing extra. Anthropic keeps its subprocessor list externally; read it before you sign.
Third, certifications: Anthropic's Privacy Center lists SOC 2 Type I and II, ISO 27001:2022 and ISO/IEC 42001:2023 for its commercial products, plus a HIPAA offering with a BAA (as of March 2026). That's the audit basis your IT security team asks for, but a certificate doesn't replace your own assessment.
Fourth, data location, and this is where it gets uncomfortable for European companies. Anthropic does publish data residency documentation, but it describes the developer platform: API requests and workspaces in the Console. There, inference knows "us" and "global", workspace location knows only "us", and US-only inference is billed at 1.1 times the price. Only part of that carries over to the web application Team and Enterprise run in: Anthropic's help centre does list "US-only inference" explicitly as an Enterprise feature. That is a commitment about processing, not about storage. Where Team and Enterprise data actually lives, I could not document publicly. If processing inside the EU is a hard requirement for you, it has to be settled with Anthropic in writing before you sign. The path on which Claude runs in EU regions without that clarification, via Amazon Bedrock with an EU inference profile, is described in Claude on AWS Bedrock: EU, GDPR and Cost.
Fifth, a detail for organizations with zero data retention: Anthropic's help center says that since 9 June 2026, for certain model classes, a 30-day retention applies even under zero-data-retention agreements. With a ZDR requirement, raise that before you sign.
Six features only Enterprise has
On its Enterprise page and in the help center, Anthropic lists features Team doesn't have. The six that decide the question in practice:
- SCIM: User accounts are created and removed automatically from your identity provider, which matters past a few dozen people.
- Audit logs: Who did what and when. Without them your compliance team has no evidence trail.
- Custom data retention: You define how long chats are kept instead of accepting the default periods.
- Compliance API: Programmatic access to usage and content data for your own audit tooling.
- Domain capture and fine-grained roles: Every account on your company domain lands in your organization automatically, and permissions split more finely than in Team.
- Customer-managed encryption keys: You hold the key, not Anthropic.
On top of that comes "US-only inference", the commitment to keep your organization's inference inside the United States, and a security product called Claude Security in beta. The context window is not a difference: the pricing page lists up to 1 million tokens for Team and Enterprise alike, varying by model. If none of these six appears in your requirements, there's little argument for Enterprise beyond paying for usage on invoice.
My decision: Team for business data
Proposals, internal documentation and receipts go through Claude here. Peter, my AI employee for bookkeeping, processed 63 receipts on 6 September 2026, asked two clarifying questions before producing output, and built thirteen quarterly folders plus an Excel file. What else he works on, from receipt types to matching the credit card statement, is on Peter, my employee for bookkeeping. Data like that contains names and addresses of business partners, and for me that belongs under a data processing agreement. That's why business data went to Team, with two seats. Max runs alongside it for my own daily work in the terminal; at times I even had two Max subscriptions in parallel while working intensively with knowledge graphs. Team for business data, Max for the daily grind: I'm very happy with that split.
That makes the Enterprise question quick to answer here, and not on price but on headcount: two seats are a long way below the 20 Anthropic requires as a minimum. I work essentially as a solo business owner with one business partner, and my AI employees aren't seats; they're folders with rules that run under my account. A folder needs no licence.
If you're under 20 people like me, the choice between Team and Enterprise doesn't exist. That's no loss, because Team already carries the privacy commitments: the training exclusion and the data processing agreement are the same ones. What I'd be missing is audit logs and custom retention. I deliberately go without both, and that is my risk decision for my own business, not a general rule: anyone who has to prove things to customers, a regulator or an auditor needs logs and retention periods even with two seats. How I built my workforce is described in AI Employees for Consultants and Agencies.
Which plan for whom: four sizes
Headcount decides first, then your IT requirements. Four typical cases, deliberately written as patterns rather than experiences:
2 to 10 people: Team Standard on an annual contract. A seat then costs the same as a Pro account monthly, and you get the DPA, the training exclusion and central administration with it. If you have heavy users, mix in Premium seats.
10 to 20 people: Still Team, because Enterprise can't be bought. If your IT asks about SCIM or audit logs, plan for Enterprise once you reach 20.
20 to 50 people: Here you can choose. Team if fixed costs matter more than administrative depth. Self-serve Enterprise if audit logs, custom retention or your own keys are in the requirements and you'll buy credits up front and maintain spend limits.
Over 50 people: Self-serve Enterprise or Enterprise through sales, which Anthropic offers from 50 seats. Through sales, usage is billed monthly in arrears and you can pay by invoice. Above 150 people Team drops out anyway.
In every case: whether Team or Enterprise, Claude Code needs no separate contract, it's in both plans. If you want Claude Code for one person only, run the numbers in Claude Code Pricing: Plan or API? first.
Frequently asked questions
How much does Claude Enterprise cost per user?
The pricing page lists 20 dollars per seat per month, billed annually, plus usage by consumption at API prices (as of October 2026). What a user really costs depends on model and tasks. Anthropic gives no guideline for total usage, only for Claude Code alone: on average 150 to 250 dollars per developer per month (code.claude.com, as of 26 October 2026).
How many employees do you need for Claude Enterprise?
At least 20 seats, according to Anthropic's help article on seat management. Self-serve online starts at 20 seats; buying through sales starts at 50 seats. Both routes have the same features and the same seat price. Below 20 people, Team is the only business plan.
Is Claude Enterprise GDPR compliant?
Only your data protection officer can answer that for your case; this is not legal advice. Documented: no training on your content by default, a Data Processing Addendum with EU Standard Contractual Clauses in the commercial terms, SOC 2 and ISO 27001. Not documented: data residency for the Claude app, because the residency docs cover the API and the Console.
Where is data stored on Claude Team?
Anthropic's data residency documentation describes the developer platform, the API and the Console, where workspace location is "us" and inference is "us" or "global". For Claude Enterprise, "US-only inference" is now documented as a feature, but that concerns processing, not storage. An equivalent public commitment on where Team and Enterprise data lives is something I did not find. If you need a storage location, ask Anthropic in writing and have the answer written into your contract.
What's the privacy difference between Claude Enterprise and Team?
None on the training exclusion and the data processing agreement; both plans sit on the same contractual level. Enterprise adds audit logs, custom retention, a Compliance API and customer-managed encryption keys. Those are tools for your evidence obligations, not different commitments on how your data is handled.
Can I switch from Team to Enterprise?
Yes. Anthropic's help center describes an upgrade path that keeps your Team organization: chats, projects, shared content, members and roles carry over. Self-serve, you start the switch online with at least 20 seats, and it can't be reversed. Alternatively, it runs through sales. After the switch some features start out turned off, such as skills, code execution and Claude in Chrome, and need to be switched back on (Anthropic help article on moving from Team to Enterprise, as of October 2026).
Where to go from here
Count your seats first. Below 20, Team is your plan, and the numbers are in Claude Team: Cost and When It Pays Off. From 20 upward, check your requirements against the six Enterprise features above and weigh usage by consumption against Team's fixed allowance. What Anthropic commits to on privacy is in Claude Privacy and GDPR: What Applies, and all prices with sources in Claude Pricing and Plans Explained. If you want Claude in your company not just as a chat but as an AI employee with its own working folder: the ready-made AI employees are available to load in my community.
