Claude Enterprise vs Team for a small company
Anthropic sells Claude to companies on two plans: Team, for 2 to 150 people, and Enterprise. [1][3] They are priced differently. An Enterprise seat covers access only, every token your people use is billed separately at API rates, and buying it online takes at least 20 seats. [1][2] So for a company of 10, 30 or 80 people the question is not which plan is cheaper. It is which controls you actually need, and whether you can buy them at your size.
Here is what Team already includes, what Enterprise adds, how Enterprise billing works, what moving from Team changes, and, at the end, a different approach, a company server, which is what we build.
Team and Enterprise at a glance
| Team | Enterprise, self-serve | Enterprise, sales-assisted | |
|---|---|---|---|
| Size | 2 to 150 seats [3] | 20-seat minimum [2] | 50-seat minimum [2] |
| Seat price | Standard $20, Premium $100 per seat a month billed annually; $25 and $125 monthly [3] | $20 per seat a month, billed annually [1] | The same list price; volume and committed-spend discounts through sales [1][2] |
| Usage | Included up to each seat’s limits; past them, usage credits at standard API rates [1][3] | Not included: every token billed at API rates from prepaid credits; no per-seat limits [2][4] | Not included: billed monthly in arrears [2][4] |
| How you pay | Credit card, monthly or annually [1] | Card, debit card or ACH, US dollars only, annually [1][2][4] | Card, ACH or invoice, other currencies, annually [1][2] |
What Team already includes
A lot of what small companies think of as “enterprise” is already in Team: single sign-on (SSO), just-in-time provisioning, spend caps for the organization and for each person, enterprise search across Slack, Microsoft 365 and custom connectors, connectors for Google Drive, Gmail, Google Calendar, GitHub, Microsoft 365 and Slack, Claude Code and Cowork, and monthly credits for the Claude API. [3] The pricing page puts those credits at up to $500 a month, pooled, and adds usage analytics, organization-wide skills deployment, admin controls for remote and local connectors, and no model training on your content by default. [1]
Same seat price, less included. On an annual plan a Standard Team seat and an Enterprise seat both cost $20 a month, but the Team seat includes usage, and people who reach their limits can keep working on usage credits at standard API rates. [1][3] For the same use, Enterprise is not the cheaper plan. What the difference buys is its controls.
If seats and usage limits are your question, our post on Claude Team plan usage limits and unused seats goes through them.
What Enterprise adds
Enterprise includes everything in Team, plus: [1][2]
- Audit logs of user actions, system events and data access, and a Compliance API that gives programmatic access to activity logs, chat histories and file content, filtered by user and time
- SCIM, so people are added and removed from your identity provider
- Custom data retention controls
- Customer-managed encryption keys: a key in your own cloud provider that Anthropic uses to protect your team’s chats, projects and files
- US-only inference, billed at 1.1 times standard API rates for Claude Opus 4.6, Claude Sonnet 4.6 and later models [4]
- Finer access control: role-based access with fine-grained permissioning, network-level access control and IP allowlisting; after an upgrade, owners can set up custom roles, groups and group spend limits [1][5]
- Domain capture, and moving existing accounts on your domains into the organization; the pricing page marks both Enterprise only [1]
- An Analytics API for aggregated engagement and adoption metrics
- A HIPAA-ready configuration with a business associate agreement, for eligible organizations
- Claude Security, in beta [1]
These are the controls an IT or compliance owner uses to show who did what, to keep data for a set period, or to handle regulated data. If nobody at your company would read an audit log or run a SCIM sync, they are part of the price rather than a reason to pay it.
How Enterprise billing works
- The seat buys access, not usage. Each seat gives a person Claude on the web, desktop and mobile, plus Claude Code and Cowork. Every token they use in any of them is billed at standard API rates on top. There is no per-seat limit and no included allowance. [2][4]
- Self-serve plans run on prepaid credits. All credits are shared across the organization; when they run out, usage stops for everyone until an owner (or a custom role allowed to manage billing) buys more. [4] Auto-reload is available with a credit card, not with ACH. [2]
- Self-serve seats are fixed for the year. Seats can’t be removed mid-term. [4]
- Sales-assisted plans are billed monthly in arrears, with no balance to run down. [4]
- Spend limits can be set for the organization and for each person. On a self-serve plan, a person who reaches their limit stops even if credits remain. [4] When the organization’s limit is reached, everyone loses access immediately until it is raised or the month turns. [5]
- No monthly API credits. Team includes them; Enterprise is not eligible. [3][5]
- Annual only. [1]
In practice, a person who barely uses Claude costs about the seat, and a heavy Claude Code user costs the seat plus everything they use. The minimum is the seat fee alone: 20 seats at $20 a month is $4,800 a year before any usage, the same as 20 Standard Team seats on an annual plan, which include usage. [1][3]
Moving from Team to Enterprise
Anthropic recommends upgrading the same organization rather than starting a new one. [5]
- What carries over: chat history, projects and shared content, memberships and roles. [5]
- The self-serve upgrade can’t be undone. You add seats for everyone, with the 20-seat minimum, and set a per-user spend limit and a starting usage balance. [5]
- Some things are off by default after the move: skills (with skill creation and sharing), code execution and file creation, Claude Design, Slides and Docs, standalone Claude Design and Claude in Chrome. Nothing is deleted; an admin turns them back on. [5]
- Before and after the move: raise any spend limit you set on Team before the go-live date, so people aren’t locked out, and afterwards look for people shown as “No seat assigned”, who have no access until an admin fixes it. [5]
- Credits: unused Team usage credits roll over; new monthly API credits stop, and the ones you already have stay usable until they expire. [5]
- Sign-in: turning on SSO logs everyone out once, and SCIM from Microsoft Entra ID syncs about every 40 minutes. [5]
When a small company needs Enterprise
Three questions settle it.
- Does someone need audit logs, SCIM, retention controls, your own encryption keys, domain capture or a HIPAA configuration? Among Claude’s plans, those come only with Enterprise. [1][2]
- Do you need more than 150 seats? That is where Team stops. [3]
- Do you have 20 people who will use it? Under 20, self-serve Enterprise means paying for empty seats, and on self-serve they can’t be removed mid-term. [2][4]
If the first two answers are no, Team is the plan, whatever the third. Enterprise is not a way to save on uneven use: its seat costs what an annual Standard Team seat costs, without the included usage. [1][3]
A different approach: a company server
Another option is to run the AI where the company’s files are, and set access and limits there. That is what we build, so read this section as ours.
Trina AX runs Claude Code, and optionally Codex, on one company server with the company’s files, the team’s skills and each person’s folder permissions; people use it through a chat app in the browser, on the desktop or on a phone. [10]
- An activity log: sign-ins, admin actions, file changes and the AI’s tool use, by day. It never shows chat content; admins can’t open other people’s personal folders or read their chats, except chats shared in a team project they add themselves to (which its members are told) or shared with them. [6][19][24]
- Per-person usage and limits: answers, tokens and an estimated cost by person, group, engine, model, day, AI connection or skill, with limits for a person, a group or the whole team, per day or per month. The costs are estimates of the API price, not bills. [7]
- Folder permissions on every file: the server checks the asking person’s access on every file request, and the engine’s own shell and file tools are turned off. [8]
- Your AI accounts: Claude or OpenAI, a subscription or an API key, for a person, a group or the team. On new teams one company connection can serve the whole team by default [20]; that software setting does not grant permission under Anthropic’s or OpenAI’s terms, and sharing one person’s subscription across a team can break them. A company API key with a monthly limit per person avoids that question. [10]
- Priced by team size, not seats: Trina hosting runs a private server for your company, in Seoul, with plans from $50 a month for up to 5 people, up to 40 on the largest listed plan and a quote above that; each plan also caps how many AI answers run at the same time. The software is free and open source (AGPL-3.0) to run on your own server. AI usage is paid separately to your AI provider. [10]
What it doesn’t do. People sign in with a username and password; the docs describe no single sign-on or SCIM. [9] Trina AX has no security certification yet, and people who operate the server or its storage can read stored files and chats, which on a hosted plan includes Trina’s operators. [8][11] Self-hosting is not offline: what the AI needs to answer goes to the AI provider you connect, and web searches, the web pages the AI opens, the services you connect, voice input and the sound of video and audio files when the speech provider is OpenRouter (on Trina hosting both go there by default and an admin can turn either off; with the team’s own OpenRouter key, file sound goes only if an admin allows it; a self-installed server can also turn speech into text on the server itself) [19], and periodic address and update checks to updates.trina.app leave the server too. [8][14] Since version 0.5.6, Find work files on this computer, in the Windows app and the Mac app from the website, sends the names, folder names, types, sizes and dates of up to 80 files on a person’s computer to that person’s AI provider when they start it (files that look personal are left out), and file contents leave the computer only for the files they add, which go to their My files on the team server. [25][26] Since version 0.5.4 (2026-11-14), by default, phone-app notifications go through Trina’s push service and then Apple or Google, on a self-installed server too, as ids, an event code, a language code and the phone’s notification token with no message text; an owner or admin can turn this off unless the server’s host has set it. [21][22] Where the dev environment is on (a sandbox in which the AI runs commands such as git or npm on the server and can read what the asking person can read; by default for new teams on servers you install yourself on Debian 13, Ubuntu 24.04 or a team server on a Mac, not the office Mac), those commands can also send data out through the sites they may reach. [23] Since version 0.5.1 (2026-11-12), file contents also leave the server for a nightly file list: each night the team’s AI account reads new and changed files in the shared, project and group folders and in folders connected from the server’s disk, a capped amount at a time, and writes one line per file, so those contents go to that account’s AI provider. It counts against that account’s usage limit. Personal folders are not read; since version 0.5.3 (2026-11-13) the list is always on, and only the server’s host can turn it off, in the server’s settings file. [18] Also on by default since 0.5.1, once a day the team’s AI account reads the first lines of the questions people typed in their own AI chats over the last 30 days, with links, emails, numbers and the names of the team’s members, projects and group chats masked, and writes a few general questions that three or more people asked in a similar way; those first lines go to that account’s AI provider too, and an admin can turn it off. [13][15] People work in Trina AX’s own chat app, not in Claude’s apps, though once the server’s operator turns it on a person can connect their own Claude or Codex to it as a remote MCP server, which then works with the team’s files and, unless they untick it, their conversations; what it reads or sends there writes no audit entry. [10][12][16] And the activity log is not a compliance archive of conversations; by design it leaves chats out. [6]
Fits when you have fewer than 20 people and want company folders with per-person access and per-person usage limits on Claude Code or Codex, without an identity provider to connect.
Side by side
| Claude Team | Claude Enterprise | Trina AX | |
|---|---|---|---|
| Smallest size | 2 seats [3] | 20 seats (self-serve) [2] | The smallest hosting plan, or your own server [10] |
| What you pay for | Seats with usage limits [3] | Seats plus all usage at API rates [2] | A server plan, or your own server; AI usage to your provider [10] |
| Sign-in | SSO, JIT [1][3] | Adds SCIM and domain capture [1][2] | Username and password [9] |
| Records | Usage analytics [1] | Audit logs, Compliance API [2] | Activity log without chat content [6] |
| Limits per person | Spend caps [3] | Spend limits for person, group, organization [4][5] | Token or dollar limits for person, group, team [7][17] |
How to choose
- Under 20 people, no compliance owner → Team. For seats and limits, see our Team plan post.
- Audit logs, SCIM, retention, your own encryption keys, domain capture or HIPAA required, or more than 150 seats → Enterprise: self-serve unless you need invoicing or another currency, sales-assisted from 50 seats.
- A small team on Claude Code or Codex with company keys, which wants folder permissions and per-person limits rather than an identity provider → a company server, like Trina AX.
Questions people ask
How much does Claude Enterprise cost?
$20 per seat a month, billed annually, plus all usage at standard API rates, with a 20-seat minimum when you buy it online. [1][2] The seat fee alone is at least $4,800 a year. [1] Sales can offer volume and committed-spend discounts. [1]
What is the minimum number of seats for Claude Enterprise?
20 for self-serve Enterprise and 50 for sales-assisted. Team runs from 2 to 150 seats. [2][3]
What does Claude Enterprise have that Team doesn’t?
Audit logs, the Compliance API, SCIM, custom data retention, customer-managed encryption keys, US-only inference, domain capture, role-based access and other access controls including IP allowlisting, and a HIPAA-ready option. Usage is billed at API rates instead of per-seat limits. [1][2]
Can I move from Team to Enterprise and back?
Upgrading the same organization keeps chats, projects, members and roles, but the self-serve upgrade can’t be reversed. [5]
Try it
- Anthropic’s plans: claude.com/pricing. [1]
- Trina AX: Set monthly usage limits, Connect a company API key and Admin. Source: github.com/trina-app/trina-ax. Hosting plans: ax.trina.app/#pricing.
Claude is a trademark of Anthropic. Trina AX is made by Trina Inc. and is not affiliated with Anthropic.
Sources
- [1] Pricing · claude.com/pricing · read 2026-11-08
- [2] What is the Enterprise plan? · support.claude.com/en/articles/9797531-what-is-the-enterprise-plan · read 2026-11-08
- [3] What is the Team plan? · support.claude.com/en/articles/9266767-what-is-the-team-plan · read 2026-11-08
- [4] How am I billed for my Enterprise plan? · support.claude.com/en/articles/11526368-how-am-i-billed-for-my-enterprise-plan · read 2026-11-08
- [5] Migrate your organization from Team to Enterprise · support.claude.com/en/articles/13779868-migrate-your-organization-from-team-to-enterprise · read 2026-11-08
- [6] Admin · ax.trina.app/docs/use/admin/ · read 2026-11-08
- [7] Usage and limits · ax.trina.app/docs/use/usage-and-limits/ · read 2026-11-08
- [8] Security and data flows · ax.trina.app/docs/operate/security/ · read 2026-11-08
- [9] HTTP API · ax.trina.app/docs/reference/http-api/ · read 2026-11-08
- [10] Trina AX for LLMs (llms.txt) · ax.trina.app/llms.txt; Trina AX: Claude Code and Codex for your whole team · ax.trina.app/ · read 2026-11-08
- [11] Security and how it works — Trina AX · ax.trina.app/security · read 2026-11-08
- [12] Team AX MCP connector · ax.trina.app/docs/agents/mcp-connector/ · read 2026-11-08
- [13] Changelog (v0.5.2) · github.com/trina-app/trina-ax/blob/v0.5.2/CHANGELOG.md · read 2026-11-08
- [14] Security and permissions (v0.5.2) · github.com/trina-app/trina-ax/blob/v0.5.2/docs/security-and-permissions.md · read 2026-11-08
- [15] server/src/chat/often-asked.ts · github.com/trina-app/trina-ax/blob/v0.5.2/server/src/chat/often-asked.ts · read 2026-11-08
- [16] Team AX MCP connector (v0.5.2) · github.com/trina-app/trina-ax/blob/v0.5.2/docs/site/agents/mcp-connector.md · read 2026-11-08
- [17] Admin (v0.5.2) · github.com/trina-app/trina-ax/blob/v0.5.2/docs/site/use/admin.md · read 2026-11-08
- [18] Changelog (v0.5.3) · github.com/trina-app/trina-ax/blob/v0.5.3/CHANGELOG.md · read 2026-11-08
- [19] Security and permissions (v0.5.3) · github.com/trina-app/trina-ax/blob/v0.5.3/docs/security-and-permissions.md · read 2026-11-08
- [20] server/src/credentials/policy.ts (v0.5.3) · github.com/trina-app/trina-ax/blob/v0.5.3/server/src/credentials/policy.ts · read 2026-11-08
- [21] Changelog (v0.5.4) · github.com/trina-app/trina-ax/blob/v0.5.4/CHANGELOG.md · read 2026-11-08
- [22] server/src/notifications/push-native.ts (v0.5.4) · github.com/trina-app/trina-ax/blob/v0.5.4/server/src/notifications/push-native.ts · read 2026-11-08
- [23] server/src/devenv/setting.ts (v0.5.4) · github.com/trina-app/trina-ax/blob/v0.5.4/server/src/devenv/setting.ts · read 2026-11-08
- [24] server/src/chat/shared.ts (v0.5.4) · github.com/trina-app/trina-ax/blob/v0.5.4/server/src/chat/shared.ts · read 2026-11-08
- [25] Find work files on this computer (v0.5.7) · github.com/trina-app/trina-ax/blob/v0.5.7/docs/site/use/find-work-files.md · read 2026-11-08
- [26] Changelog (v0.5.7) · github.com/trina-app/trina-ax/blob/v0.5.7/CHANGELOG.md · read 2026-11-08