What is the difference between Claude Team and Enterprise?

What each plan costs, and what the seat fee actually buys

The two plans are priced on completely different models, and that is the part that wrecks budgets. Team bundles usage into the seat. Enterprise sells you access and then bills every token on top at API rates. A finance team that models Enterprise as a seat count times a seat price will be wrong, sometimes by a lot.

Anthropic's support article for Team publishes exact numbers: a minimum of two members, a ceiling of 150 seats, and per member per month pricing of 25 dollars monthly or 20 dollars annually for a Standard seat and 125 dollars monthly or 100 dollars annually for a Premium seat. Those are US prices before tax, and Anthropic says pricing, currency, and tax handling vary by region.

Enterprise is thinner on published numbers. The plans page on claude.com carries the figure: one Enterprise seat type at 20 dollars per seat per month plus usage billed at API rates, billed annually. The support article for Enterprise deliberately does not state the price and instead points you to a pricing page URL that redirects to a solutions page with no price on it. So the number lives on the plans page and nowhere else in the documentation. Read it live before you put it in a budget, because Anthropic states outright that price and plans are subject to change at its discretion.

The usage difference is worth spelling out for whoever owns the forecast. On Team, each member gets their own included allowance, not a shared pool, and Anthropic quantifies it: a Standard seat is 1.25x the usage of a Pro plan per five hour session, a Premium seat is 6.25x, with weekly limits on top and an option to buy usage credits when someone runs out. On Enterprise there is no seat allowance at all to run out of, which also means there is no ceiling. Spend limits at the organization and per user level are the only brake, and Anthropic warns that when the org level spend limit is hit, every user loses access immediately until it is raised or the month rolls over.

• ·Team: minimum 2 members, maximum 150 seats, monthly or annual billing, credit card only
• ·Team Standard seat: 25 dollars per member per month billed monthly, 20 dollars billed annually, usage included
• ·Team Premium seat: 125 dollars per member per month billed monthly, 100 dollars billed annually, same features with more usage
• ·Team lets you mix Standard and Premium seats in one organization and reassign people between them without buying a new seat if one is free
• ·Enterprise: one seat type, listed at 20 dollars per seat per month on the plans page, annual billing only, seat fee covers access with zero included usage
• ·Enterprise usage: every token in chat, Claude Code, and Cowork is billed at standard API rates, with no per seat limit and no included allowance
• ·Enterprise self-serve: 20 seat minimum, credit card or ACH, USD only, usage credits purchased up front
• ·Enterprise sales-assisted: 50 seat minimum, adds invoicing and net terms, multi-currency, and usage billed monthly in arrears

The capabilities that are Enterprise only

This is the section to hand to your IT or legal reviewer, because it is the actual answer to which line item you need. Team is a real commercial plan with real administration. It is not a compliance plan.

Team already covers the basics most operators assume they need: single sign-on, domain verification, just-in-time provisioning, central billing and administration, admin controls over remote and local connectors, enterprise deployment of the desktop app, usage analytics, organization-wide skills deployment, organization instructions, spend controls, and enterprise search across connected tools. For a firm whose security review is a questionnaire about SSO and who can see what, Team clears it.

Everything in the list below is Enterprise only. If your reviewer asks for any single one of these, the conversation about Team is over and you should price Enterprise instead of trying to negotiate the feature onto Team.

Two of these deserve a note. Audit logs are exportable as a CSV covering the past 180 days, and Anthropic is explicit that chat and project titles and content are not in that export, only identifiers. If you enable customer-managed encryption keys, the export button stops working entirely and you have to read audit events through the Compliance API instead. So if your answer to a security questionnaire depends on log retention beyond 180 days or on machine-readable delivery into a SIEM, plan on the Compliance API from the start rather than the CSV.

• ·SCIM provisioning and directory sync from Okta, Entra ID, or another provider
• ·Audit logs, including the CSV export of the past 180 days of organization events
• ·The Compliance API, which returns per event activity records plus chat, file, project, Cowork, and Claude Code session content, and can delete content on demand
• ·Custom data retention controls, with a 30 day minimum retention period
• ·Customer-managed encryption keys held in your own cloud provider
• ·US-only inference for your organization
• ·Role-based access using custom roles and groups, including delegating billing or user management without handing out the Owner role, though Anthropic's Team article describes role-based permissioning as a Team feature and the next section covers that conflict
• ·The HIPAA-ready configuration and the click-to-accept BAA, which Anthropic states Team and the individual plans cannot enable
• ·IP allowlisting and network-level access control through tenant restrictions
• ·Inference hooks, a beta that lets your own security server see and deny a prompt before inference
• ·The Analytics API for aggregated engagement and adoption metrics
• ·Migrating existing accounts on your domains into the organization, and domain capture, which the plans page marks Enterprise only while the Team article lists it under Team, another conflict covered below
• ·A 500k context window on the default model, where Team is listed at 200k

Where Anthropic's own documentation disagrees with itself

Two capabilities are described inconsistently across Anthropic's own pages, and both are the kind of thing a procurement reviewer will circle. Do not resolve these from a vendor page, including this one. Get the answer in writing from your Anthropic contact or verify it in the product before you sign.

First, role-based access. The support article titled What is the Team plan lists Role-based permissioning under advanced identity and access management as an included Team feature. The comparison table on the plans page marks Role-based access as unavailable on Team, the Enterprise plan card lists role-based access with fine-grained permissioning as an Enterprise addition, and the setup guide is titled Set up role-based permissions on Enterprise plans and requires an Enterprise organization. The reading that fits all four is that Team has fixed built-in roles, which the seat management article confirms as User, Admin, Owner, and Primary Owner, while custom roles, groups, and delegated admin permissions are Enterprise. The Team article's wording does not make that distinction, and if your access control requirement is anything more than three fixed tiers, assume you need Enterprise.

Second, domain capture. The Team article lists Single-Sign-On (SSO) and Domain Capture together as a Team feature. The plans page splits them into two rows, gives Team domain verification, and marks domain capture as Enterprise only alongside migrating accounts using your domains. Those are different capabilities: verifying you control a domain is not the same as automatically pulling every account on that domain into your organization. If your rollout plan depends on sweeping up staff who already signed up with a personal Claude account on your company email, confirm which one you are actually getting.

How data handling differs across consumer, Team, and Enterprise

This is the detail most people get wrong, and the mistake usually runs in the reassuring direction. Free, Pro, and Max are consumer products governed by the Consumer Terms of Service. Team and Enterprise are commercial products, which Anthropic groups together as Claude for Work, governed by the commercial agreement between Anthropic and your organization.

On the commercial side the training commitment is contractual, not a setting. Anthropic's Commercial Terms state that Anthropic may not train models on Customer Content from the Services, and the privacy center article for commercial products says that by default Anthropic will not use your inputs or outputs from commercial products to train its models. The plans page comparison table matches this, listing model training as none by default for Team, Enterprise self-serve, and Enterprise sales-assisted alike. On this specific point Team and Enterprise are equivalent, and paying for Enterprise buys you nothing extra.

On the consumer side it is a per person choice, and Anthropic's own two descriptions of it read differently. The consumer privacy article says Anthropic will use your chats and coding sessions to improve its models if you choose to allow it, framing it as something you turn on. The comparison table on the plans page labels model training on Free, Pro, and Max as Opt-out, which reads as something already running that you have to turn off. Either way the practical conclusion for a business is the same and it is the important one: on a consumer plan that decision belongs to the individual employee, you have no administrative visibility into what they chose, and you cannot change it for them.

There is one exception that applies on every plan, including Enterprise. If a user clicks the thumbs up or thumbs down button, Anthropic stores the entire related conversation for up to five years and may use it to train models, de-linked from user and customer IDs. On Team and Enterprise, a Primary Owner or Owner can switch this off with the Rate chats setting under Organization settings, Data and privacy. That toggle should be on your rollout checklist, because nothing else in the default configuration sends conversation content into training.

Retention is where the two commercial plans genuinely part ways, and not in the direction most write-ups claim. Anthropic's article on custom data retention states plainly that by default data is retained indefinitely unless a custom retention period is set, and custom retention controls are an Enterprise feature. That means a Team organization has no retention dial at all. If your records policy says chat content containing client information gets purged on a schedule, Team cannot implement it and Enterprise can, with a 30 day floor.

• ·Free, Pro, Max: Consumer Terms, no central administration, training governed by each individual's own setting, no BAA path
• ·Team: commercial terms, no training by default, SSO and central billing, no retention controls, no audit logs, no BAA
• ·Enterprise: commercial terms, no training by default, plus retention controls, audit logs, the Compliance API, CMEK, US-only inference, and the BAA
• ·Every plan: the thumbs up and thumbs down feedback button stores the conversation for up to five years and may be used for training, and only Team and Enterprise owners can disable it
• ·On Team and Enterprise the Primary Owner controls the data, can export conversations and uploaded files, and can remove a member's access, and Anthropic tells users to ask that Primary Owner about the governing agreement

Which one to ask for, and why headcount is the wrong input

Before you compare plans, count the right number. Seats are not employees. At a service business of 250 to 800 people, the population that needs a Claude seat is the roles whose output is written or analytical: operations, administration, billing and revenue cycle, intake and scheduling, marketing, client communications, HR, finance, and whoever runs your systems. Field crews, clinical floor staff, drivers, technicians, and hourly production staff usually get none, at least in the first year. A 400 person company can easily land at 30 to 60 seats. Build that list role by role before you look at pricing, because it changes the arithmetic and it changes which plan is even in scope.

That has a direct consequence for the 150 seat ceiling on Team. It is a seat ceiling, not a company size ceiling. A 600 person firm rolling Claude out to 45 back office staff is nowhere near it, while a 90 person agency where nearly everyone writes for a living could brush against it. Treat the cap as a constraint on your seat plan, not as a signal about whether your company is big enough for Enterprise.

So drive the decision off requirements rather than headcount. Write down what your security reviewer, your legal reviewer, and your records retention policy actually require, then check that list against the Enterprise-only capabilities above. If nothing on your list appears there, Team is the correct plan at any company size up to its seat ceiling. If even one item appears there, Enterprise is the correct plan even if you are only buying 25 seats, because no configuration of Team produces those controls. Seat minimums then tell you which Enterprise path is open to you, not whether you need Enterprise at all.

Start with the smallest plan your reviewer will actually sign off on, because the upgrade path is clean and the downgrade path does not exist. Anthropic supports upgrading an existing Team organization in place, preserving chat history, projects, and user memberships and roles, and it warns that creating a fresh Enterprise organization instead means setting everything up from scratch. It also states that the self-serve Team to Enterprise upgrade is not reversible.

When Team is the answer, buy monthly for the first quarter so you can right-size the seat count before committing to annual, and start most people on Standard seats with a small number of Premium seats for whoever actually lives in the tool. You will find out fast who those people are, and Anthropic lets you move someone between seat types without buying a new seat when one is free. When Enterprise is the answer, budget it as seats plus a usage estimate rather than seats alone, and set per user spend limits on day one rather than after the first invoice.

Go to sales-assisted Enterprise, at the 50 seat minimum, when the blocker is commercial rather than technical: your finance team requires invoicing and net terms, you need a currency other than USD, or you want AWS Marketplace procurement. Anthropic's comparison table also puts trials, consultation, tiered incentives on committed spend, and customer success support at certain spend thresholds on the sales-assisted side only. Anthropic states that both paths include the same features and the same seat pricing, so the difference is how you buy, pay, and get supported, not what the product does. Worth knowing if you handle protected health information: the HIPAA-ready configuration is listed as available on both paths, so needing a BAA does not by itself force you into a sales cycle.

One rollout detail worth planning around, because it surprises people mid-migration. After a Team to Enterprise upgrade, several capabilities that were on by default under Team arrive switched off: skills including skill creation and sharing, code execution and file creation, interactive content in artifacts, Claude Design, and Claude in Chrome. The underlying content is preserved and turning the setting back on restores it, but if you migrate on a Tuesday and say nothing, your team will think the product broke. Raise your Team spend limit before the go-live date, verify seat assignments after cutover because some members can land with no seat assigned, and if you are going sales-assisted, schedule the cutover so users log back in the next business morning rather than mid-afternoon.

Finally, the option nobody should choose. Leaving staff on personal Pro accounts and reimbursing them is the most expensive plan in the room once something goes wrong. No central billing, no SSO, no audit trail, no BAA path, no retention policy, and a training setting that belongs to each employee rather than to you. If Claude is doing work that touches client information, the cheapest correct answer is Team and the correct answer under a compliance review is Enterprise.

• ·Count seats by role, not by employee: operations, administration, billing, intake and scheduling, marketing, client communications, HR, finance, and IT
• ·Field, clinical, driving, technician, and hourly floor roles usually need no seat in a first rollout, which is why a 400 person firm often buys 30 to 60
• ·Choose Enterprise if your reviewer requires any of SCIM, audit logs, a defined retention period, the Compliance API, customer-managed encryption keys, US-only inference, IP allowlisting, custom roles and groups, or a BAA
• ·Choose Team if the requirement list stops at SSO, central billing, admin control over connectors, spend limits, and usage analytics, no matter how many people the company employs
• ·Enterprise seat minimums gate the purchasing path, not the decision: 20 seats self-serve, 50 seats sales-assisted
• ·Team's 150 seat ceiling constrains your seat plan, not your company size

What people ask next

• How do I roll out Claude to my whole team?
• How much does Claude cost for a team?
• Will Claude pass our IT security review?
• We already have Microsoft Copilot. Why would we add Claude?

Worked examples