Claude Enterprise: What IT Gets Beyond Team
What Claude Enterprise adds to Team: identity, audit, retention, usage and connector controls, plus when to upgrade and what to ask sales.
Published Updated
Claude
- SSkills

Claude Enterprise earns its place when you need automated user provisioning, audit access or custom retention. For a rollout of up to 150 licensed users, Claude Team already includes single sign-on and Claude Code; choose between them by the controls your company requires and the number of people who need access.
Since October 28, 2026, smart reports, in-app surveys and pooled group budgets have been generally available for Claude Enterprise, adding ways to review work, collect employee feedback and allocate shared budgets. Anthropic's announcement.
Verified against Anthropic's live pricing, Enterprise page and help-center documentation on 3 November 2026. The comparison below describes the current offer for new customers. Existing Enterprise contracts can still have older seat types until renewal. Team limits, Enterprise plans.
Claude Team vs Enterprise: The Operational Difference
Claude Team is already a managed workplace subscription. Anthropic's Claude Enterprise plan adds more control over access, records and data lifecycle, alongside a different consumption model.
Think of Team as an office with employee badges and a shared administrator. Enterprise adds a directory-controlled badge desk, an activity record and configurable rules for how long documents stay. The extra controls matter when someone must operate and verify them.
Both official sales pages publish that Enterprise base price and separate usage charge. The tailored commercial offer is where you ask for a sales quote: the pricing page invites one and lists committed-spend incentives and support at certain thresholds without publishing those amounts. Do not turn those omissions into an assumed discount or fixed all-in bill.
What the Claude Enterprise Plan Adds
Enterprise becomes useful when access must follow company policy automatically and your administrators need evidence that the policy worked. Buying the plan creates the capability; configuring it makes the capability useful.
Since October 23, 2026, unified Claude has been generally available for Enterprise: chat and Cowork share a conversation across web, desktop and mobile. Work can continue in Anthropic's cloud after a laptop closes. IT now needs to govern agentic work alongside chat, including model access, spending and permitted network destinations. Enterprise announcement.
Claude Enterprise Features: Identity and Delegated Administration
SSO is shared; SCIM is the upgrade. Single sign-on lets employees use their company identity to authenticate. Just-in-time provisioning creates their Claude membership when they first sign in. SCIM, a standard for synchronizing user accounts, connects membership management to your identity provider. Anthropic makes JIT available on Team and Enterprise, while SCIM requires Enterprise for a Claude workplace organization. Identity setup.
For an operations lead handling frequent joiners and leavers, the relevant question is who owns account removal and how you verify it. SSO answers how someone signs in; directory synchronization addresses the account lifecycle. Test an employee joining, changing groups and leaving before a broad rollout.
Enterprise custom roles also let you delegate areas such as billing or identity administration and control access to features. They work through groups and apply to members assigned the Custom role. A role cannot enable a feature disabled for the organization. Role behavior.
A mid-market CTO can therefore separate engineering access from an operations group's needs. The administrative job is to define those groups and verify effective permissions, especially for employees who belong to several groups.
Audit Controls Give Your Security Team Something to Investigate
Auditability means being able to retrieve evidence of activity. Enterprise's Compliance API, a programmatic interface for your monitoring tools, can pull activity events, chat data and file content; it also includes audit-log events. Only the Primary Owner can enable it in Organization settings > API. Compliance API access.
For a security lead, this creates a concrete acceptance test: retrieve an expected event, identify who can read the resulting records, and confirm how your own monitoring system handles them. A dashboard showing adoption does not answer the same question as an investigation trail.
Before purchase, name the person responsible for the integration. An unused audit feature will not produce the review process your customers or internal policies expect.
Claude Enterprise Data Privacy: Training and Retention Are Separate
Team and Enterprise both exclude commercial inputs and outputs from model training by default. Anthropic's published policy allows training when you explicitly submit feedback or bugs, or otherwise opt in. Owners can disable thumbs-up/down feedback through Rate chats under Organization settings > Data and Privacy. Keep the words “by default” in your internal explanation. Commercial training policy.
Training describes whether data helps improve models. Retention describes how long stored content remains. Buying Enterprise for its retention controls does not mean Team content was being used for training.
Enterprise custom retention has a 30-day minimum. Standalone chats use the latest activity in the conversation or a cloud session started from it; projects use their latest qualifying activity. Data is retained indefinitely unless a custom period is configured. Most consequentially, a project’s retention policy governs its chats, even when you have configured chat retention but left project retention unset. Retention rules.
For example, an operator can set a chat retention period and still retain an old conversation inside an active project. Review both settings before telling employees that old conversations disappear on a single company-wide schedule.

Retention coverage also depends on the feature. Standalone Claude Design, Claude Tag, Claude Managed Agents and other Claude Code-in-the-cloud features remain outside the general chat/project policy. Artifacts have their own retention rules; public-link artifacts are excluded from those periods. General scope.
In unified Claude, deleting a conversation also deletes the cloud sessions it started and their transcripts. Account memory follows chat retention, and project memory follows project deletion. Independently started Cowork cloud sessions and scheduled tasks and runs remain outside this coverage; a scheduled task can keep running after its originating conversation is deleted. Ask for a retention map covering the exact features you will enable. Unified-experience scope.
Usage Limits: Capacity and Spending Are Different Controls
Current usage-based Enterprise removes the plan-level and seat-level usage allowances described for Team. It does not make consumption free or guarantee that work can never stop. Enterprise usage model.
The buying route changes the operational failure mode:
- Self-serve Enterprise: everyone draws from a shared prepaid credit balance. When that balance runs out, usage stops across the organization.
- Sales-assisted Enterprise: consumption is billed monthly in arrears, meaning after it occurs. There is no prepaid balance to exhaust.
- Both: spend limits can stop work even when the billing route would otherwise allow it. Billing and spend controls.
For unified Claude, use the four controls together:
- Efficient mode: restricts unified chat and agentic work to Sonnet and Haiku. Owners enable it on the organization's Models page or a custom role's Models tab. New sessions start on Sonnet 5.5 at low effort; other products are unaffected. Efficient mode.
- Model entitlements by role: set which models a role can use on its Models tab. Organization restrictions remain the ceiling, and permissions combine across a member's roles. Check overlapping memberships before promising that a group cannot use a model. Model access.
- Organization, group and member spend limits: configure them under Organization settings > Usage. A group limit supplies a monthly limit for each member; an individual override takes precedence over that group-derived limit. This is different from a shared departmental pot. Group limits.
- Pooled group budgets, now generally available: give a group one shared monthly limit, with individual limits kept as guardrails against unusually high use. Each request counts against the member limit and the applicable pool. A member stops when their own limit is reached or every pool they can draw from is exhausted. Dividing the pool equally into individual caps defeats the benefit of sharing it. Pooled budgets.
The help center lists pooled budgets for usage-based Enterprise plans and Standard or Premium seat plans with usage credits enabled. On Standard and Premium seat plans, only usage credits draw from the pool; exhausting it does not remove the seat's included usage. Plan eligibility.
Primary Owners, Owners and custom roles with Billing set to Can manage configure pools in Organization settings > Usage > Spend limits > By group/tier. Choose Edit limit and budget, then set Monthly pooled budget and Member monthly limit together. Pooled budgets are managed in Organization settings, and are not available through the Admin API. Setup and limitations.
When a member reaches a limit, Request more appears inside Claude; administrators find those spend requests under Review requests. Everyone with Billing permission receives email alerts when a pool reaches 50%, 75%, 95% and 100%, with notices in Organization settings and on the Usage page. Assign someone to act on those alerts before a busy group runs out. Requests and alerts.
Pricing remains usage-based. Anthropic says Opus 5.5 costs 40% less to run than Opus 5 on typical workloads, and Sonnet 5.5 costs up to 30% less than Sonnet 5 for most work. These are Anthropic's workload-efficiency claims, not guaranteed savings on your bill or a seat-price reduction. Announcement.
For a founder giving engineering heavy coding workloads and operations lighter document work, moving to Enterprise changes what finance must watch. The seat count tells you who has access; actual consumption tells you the variable bill. Give the pilot group a shared budget and assign an owner for reviewing usage and resolving blocked work.
A Team organization can instead mix Standard and Premium seats and enable usage credits after included limits are reached. A heavy user alone does not establish a need for Enterprise governance. Team options.
Does Claude Enterprise Include Claude Code?
Yes, the current Enterprise seat includes Claude Code, Anthropic's coding tool for terminals and supported development environments. Every Team seat includes it too. Sign in through Claude account with subscription, then select the work organization. Access is included; the plan's consumption rules still apply. Claude Code setup.
Existing Enterprise accounts may show legacy Chat, Chat + Claude Code, Standard or Premium seats. Check the seat type before applying the new-customer comparison. Anthropic says those older arrangements transition at renewal. For the broader coding subscription choices, see Claude Code pricing.
Connectors: Buy the Permission Model Your Work Requires
Workplace integrations are available on Team as well as Enterprise. Anthropic lists Google Drive, Gmail, Google Calendar, GitHub, Microsoft 365 and Slack; enterprise search is also available on Team. Team integrations.
Unified Claude's agentic work runs in Anthropic-managed cloud sessions, each in its own isolated virtual machine. Network egress is off by default for Enterprise, and connector credentials stay outside the sandbox. Cloud architecture. Egress means outbound network access from code in that environment. In Organization settings > Capabilities > Code execution, admins can leave it off, allow package managers, specify domains or allow all domains. Connector policies are enforced separately, outside the sandbox, on every call. Review both controls before approving a workflow that fetches external data. Network and connector settings.
The distinction for IT is who can reach a connector and which actions they can use. Enterprise custom roles can grant, block or require approval for connector tools. Across a person's roles, grants are additive: blocking a connector in one role does not cancel another role's permission. The organization-wide tool policy remains the ceiling. Connector permissions.
For an operations lead rolling out internal search, use a low-privilege test account as well as an administrator account. Anthropic says search respects source-system permissions, with an important exception: a custom connector using shared credentials exposes what those credentials can access. Search permissions.
Also inventory locally configured connectors. The custom-role guide explicitly excludes connectors a member runs locally from those organization connector permissions. If your decision is primarily about working within Microsoft's applications, Claude vs Copilot addresses that workflow choice.
Who Should Stay on Team?
Stay on Team when its licensed-user ceiling fits, SSO and central administration meet your requirements, and you do not need SCIM, Enterprise audit access or custom retention. Team supports up to 150 seats; count the people receiving licenses, rather than everyone on payroll. Team capacity.
A hypothetical company with 300 employees and an 80-person pilot can still fit within that capacity. A company with 20 employees may need Enterprise immediately if automated provisioning or a required retention policy is a condition of adoption.

Act now when a mandatory control is missing or the planned rollout exceeds Team's seat ceiling. Stay and measure when the problem is occasional usage pressure that Team seat allocation or usage credits could address. Wait on broader deployment when nobody owns permissions, consumption review or retention configuration.
For someone evaluating Claude individually before proposing a workplace rollout, Is Claude free? covers that earlier decision.
If ChatGPT Enterprise Is Also on the Shortlist
OpenAI's ChatGPT Enterprise should face the same acceptance checklist as Claude Enterprise. Ask both vendors to demonstrate the required identity lifecycle, records access, retention scope and connector permissions against your intended configuration. Compare written commercial offers against the same expected adoption and workload, rather than comparing a published seat fee with a bundled quote.
If you already operate an approved deployment, adding another platform needs a specific workflow benefit large enough to justify another set of policies and administrators. This is a procurement decision rule, not a claim that the vendors publish identical terms. ChatGPT for business covers the adjacent workplace decision.
How to Buy and Roll Out Claude Enterprise
Choose the procurement route first, then configure the organization around a documented pilot. A sales conversation is available, but the published Enterprise offer also has an online purchase path.
Claude Enterprise Minimum Seats
Anthropic's help center specifies 20 seats for self-serve and 50 for sales-assisted Enterprise, with the same features and base seat pricing. Self-serve supports USD payment by card or ACH bank transfer. Invoicing and non-USD payment are reasons to contact sales; the pricing page also places AWS Marketplace purchasing on the sales-assisted route. Purchase routes, Procurement comparison.
Choose a new organization or upgrade the existing one
New customers can follow Get Enterprise plan on Anthropic's Enterprise page. Existing Team customers can use the upgrade flow and select Upgrade an existing organization to preserve conversations, projects and memberships. Anthropic states that the self-serve Team-to-Enterprise upgrade is irreversible. Migration guide.
Approve the access and consumption commitments
Review seats, the annual term, starting usage balance and spend limits before confirming. Self-serve Enterprise seats cannot be removed mid-term, according to the billing guide. Use Contact sales for a tailored offer or procurement requirements the online route does not meet. Billing rules.
Configure and verify before inviting everyone
Name the Primary Owner, verify provisioning, assign roles, set chat and project retention, and approve the pilot's connectors. After a Team migration, check feature settings and seat assignments: some capabilities may be off and some users may need a seat assigned. Post-migration checks.
For unified Claude, enable Chat and Cloud code execution and file creation, then Organization settings > Capabilities > Chat and Cowork unified. Separate Cowork toggles are not prerequisites. Standard User, Admin and Owner roles receive access when the organization switch is on; for a restricted pilot, put excluded members on appropriately restricted custom roles first. Review egress, model access and budgets before enabling it. Rollout settings.
Organizations using zero data retention for Claude Code, a HIPAA-ready configuration or customer-managed encryption keys stay on their current setup and cannot enable the unified experience. For eligible existing organizations that have not chosen a setting, the admin guide schedules automatic enablement for November 16, 2026, once the required capabilities are on; select Keep disabled if you need to defer. Eligibility and rollout timing.
Measure the pilot's work and remove friction
Smart reports show what the group produces and where work gets stuck. They sample transcripts for your chosen team, dates and products, then group sessions into workstreams with spend at list price, deliverables and friction points such as missing connectors or failed tool calls. A Primary Owner or Owner must first enable them under Organization settings > Capabilities > Analytics. Primary Owners, Owners, Admins and custom roles with Analytics view access can create and open reports in Analytics > Smart reports. Delegated team leads can also run reports within their assigned scope. Smart reports guide.
Smart reports exclude organizations using customer-managed encryption keys, a HIPAA configuration or hybrid inference on their own cloud provider. With zero data retention for Claude Code, reports can cover chat and Cowork but leave out Code. Anthropic specifies adoption and investment planning as their purpose, not individual performance evaluation or employment decisions. Report eligibility and purpose.
In-app surveys add employees' account of the value and obstacles. Primary Owners, Owners, Admins and custom roles with Analytics view access can create and view surveys in Analytics > Surveys, choosing the audience, dates, questions and where they appear. Employees see a small card after a chat reply, between Cowork tasks or after Claude Code finishes responding; it does not interrupt a running task. Surveys are unavailable for HIPAA configurations. They support customer-managed encryption keys, though survey data from about the first day after adding a key stays encrypted with Anthropic's key. Answers go to your organization's admins, and CSV exports include respondents' email addresses. Survey setup and eligibility.
For an IT or ops lead, pair the report with that feedback before expanding access: fix a recurring connector problem, share a useful workflow, then review the group's budget and spend requests. This gives the rollout owner a concrete action beyond counting active seats.
Questions to Ask Anthropic Sales
- Which published purchasing route and seat minimum apply to our proposed rollout, and what is the complete written quote?
- Which usage charges, committed-spend incentives and support conditions apply to that offer, and who will own group budgets?
- Which events and content can our monitoring system retrieve, and who must enable access?
- Which enabled features follow our chat/project retention policy, and which require separate handling?
- Can you demonstrate provisioning, removal and connector restrictions for our proposed groups, including members with overlapping roles?
- If we are migrating or renewing, what changes in seat types, billing and configuration, and when?
Ask for answers attached to your proposed configuration. Do not treat a feature list as a promise about unpublished contract terms.
What's Overhyped About Enterprise?
Enterprise is not a prerequisite for SSO, Code or workplace search. Those capabilities already appear in Team's published offering. The upgrade needs a more specific reason.
“No usage limit” is not a fixed-price promise. Consumption charges, spend caps and a self-serve credit balance still matter.
“Custom retention” is not a universal deletion switch. Project precedence and feature-specific exclusions require configuration and review.
The practical move next week: have the IT owner write down the missing control that would justify Enterprise, then have the pilot owner demonstrate it with a joining user, a departing user, a restricted connector and a retention configuration review. For unified Claude, also verify outbound network access, role-specific model access and the pilot's pooled budget before expanding access. If there is no missing control and Team fits the rollout, keep the simpler plan while you measure adoption.
Frequently Asked Questions
How much is a Claude Code Enterprise license?
Claude Code is included in the current Enterprise seat; consumption is charged separately. The published seat fee and billing interval are in the comparison table above. A tailored commercial package needs a sales quote. Current offer.
Which enterprises are using Claude?
Anthropic's Enterprise page lists organizations including Uber, Spotify and GitHub. That establishes the vendor's named customer examples, not which exact subscription or contract each organization has. Enterprise page.
What are the different tiers of Anthropic AI?
The pricing page lists Free, Pro, Max, Ultra, Team and Enterprise for Claude, alongside separate Claude Platform API pricing. For this workplace decision, compare Team's included allowances with Enterprise's controls and metered usage. Plan list.
How to use Claude enterprise?
Purchase online or work with sales, configure your organization, then invite employees to use their work accounts. Existing Team customers should review the in-place upgrade route before creating another organization. Migration instructions.
What do I get with Claude Enterprise?
The current seat provides access to Claude's workplace tools, including Claude Code, plus Enterprise administration such as SCIM, audit access and custom retention. Eligible organizations can enable unified chat and Cowork with role-based model access and spending controls. Usage is billed separately from access. Plan overview, Unified administration.
Can you use Claude Enterprise for personal use?
It is sold as an organizational subscription with a seat minimum, not as a single-user personal plan. An individual evaluating Claude should start with the personal options on the pricing page. Published plans.
What are the key differences between Claude Enterprise and Claude Pro?
Pro is an individual subscription. Enterprise supplies organization administration and governance, and charges for consumption separately. Team is the intermediate workplace option to evaluate before moving to Enterprise. Plan comparison.
Does Claude Enterprise have a usage limit?
Current usage-based plans have no plan-level or seat-level usage allowance cap. Spend limits can still stop work, and self-serve organizations stop when shared credits run out. Legacy seat-based Enterprise plans have different rules. Usage billing.
Does Claude Enterprise keep your data private?
Anthropic does not train on commercial inputs and outputs by default. That statement does not mean data is never stored or inaccessible to your organization: retention and authorized compliance access are separate controls. Explicit feedback or opting in can change training use. Training policy.
For practical briefings on adopting AI at work, join the newsletter.
- Last Updated
- Category
- AI
- Language








