Buying Claude Team gets every person on your team their own Claude. It is a good deal per seat, and the chats stay private to whoever opened them. Type buys your team one shared workspace that Claude works inside, where a thread is visible to the Space and a coworker can pick it up.
14-day free trial. Runs on Claude and Codex.
The Claude Team plan is Anthropic's seat-based subscription for groups of 2 to 150 people. Standard seats are $25 per member per month, or $20 billed annually, and premium seats are $125 or $100. Every seat carries Claude on web, desktop, and mobile, Claude Code, Cowork, projects, a 200k context window, SSO, JIT provisioning, and connectors an admin enables. At $20 a seat that is a lot of product, and Type does not claim otherwise.
What the plan does not buy is shared work. Anthropic states that even if a project is public, your chats within that project will be private and inaccessible to other members of your organization unless you manually share them, and that chats within a project are not shared by default. Manual sharing produces a snapshot of the messages sent before you shared it. Team plans do not have organization-level controls for memory features, and Team members are still on the legacy memory experience while Pro and Max users migrate to the new one. Connectors are only available in private projects, and chats with synced content can't be shared at all.
So the shape you are buying is N private workspaces that share a knowledge base. Type is the other shape: plans include seats for 2 to 60 team members in one workspace where the agent works in a channel, threads are visible to the Space, and memory belongs to the team. Type runs Claude and Codex, so this is not a trade between collaboration and model quality. Plenty of teams keep Claude seats for the people who live in Claude Code and use Type for the work more than one person touches.
What decides it
Both products run capable Claude agents. These are the structural choices underneath that.
It is visible to the Space. Two people prompt into the same thread, read the agent's reasoning and tool calls, and continue each other's work without asking.
Private by default, even in a public project. Anthropic states chats within a project are not shared by default, and manual sharing hands over a snapshot of the messages sent before you shared.
In the workspace and the Space. Memory, skills, and connections belong to the team, so one person's correction shows up in the next person's thread and in every scheduled run.
In each person's account. Anthropic states Team plans do not have organization-level controls for memory features and that members manage their own memory settings directly.
Type plans include 2, 4, 20, 40, or 60 team members. Everyone included in the plan can read and continue the same shared threads.
$25 a month, or $20 annually, per person, with a two-member floor and a 150-seat ceiling. Everyone who needs to see the work needs a seat of their own.
The seam
Start with what the plan does well, because at $20 a seat annually it is a lot. Claude on web, desktop, and mobile. Claude Code on every seat, terminal and IDE, with premium seats getting more usage rather than exclusive access. Cowork. Projects with a shared knowledge base. Enterprise search across your connectors. SSO and just-in-time provisioning. Usage limits that are per-member, so a colleague burning through a migration on Thursday cannot slow you down on Friday. That last one matters more than people expect, and Anthropic says it plainly: if one team member reaches their seat's included usage limit, it does not affect the limits of other team members.
Then read the sharing article. Even if a project is public, your chats within that project will be private and inaccessible to other members of your organization unless you manually share them. And: chats within a project are not shared by default. And, on sharing a private project with a colleague: this will share the project and knowledge base with the member, but your chats will remain private unless you choose to share them. What the plan shares is the reference material. The work stays with whoever did it.
Even when you do share, you share a copy. Anthropic documents that the chat snapshot includes all messages that were sent prior to sharing the chat, so the next three turns you take are invisible until you share again. Nobody prompts into your conversation. There is no version of a Claude Team plan where two people build one answer together in one chat.
Type starts one layer up. The unit of work is a thread inside a Space, and a Space is a shared computer for a team with its own channels, connections, skills, memory, and automations. When an agent works in a thread, everyone with access is already looking at it. A message can be addressed to teammates or to the agent, so the argument about what to do next sits in the same place as the work. Nothing had to be shared, because it was never private.
Memory
Anthropic is specific, so we will be too. Under the heading Team plan limitations: Team plans do not have organization-level controls for memory features. Individual Team plan members manage their own memory settings directly. There is no shared memory, and no admin lever over the memory that does exist.
Team plans are also behind. Anthropic writes that users on free, Pro and Max plans will be migrated to the new experience and that in the interim, users on Team and Enterprise plans will stay on the legacy experience. So the paying business plan runs the older memory system while individual consumers get the new one. Inside projects it fragments further: each project has its own separate memory space.
The practical version of this is the correction you made on Tuesday. You explained that your fiscal quarters start in February and that the churn number in that dashboard excludes trials. That lives in your account, on the legacy system, in one project. Your coworker running the same analysis on Thursday gets the same wrong answer you got and fixes it the same way you did.
Type memory is scoped to the workspace, the Space, or the person, generated in the background rather than curated by hand, and scheduled automations read the same memory a person's thread does. Skills go further. A skill is a procedure the team keeps improving, with an owner and version history, where a coworker's edit arrives as a change request the owner approves. Each one reports 30-day usage, top users, and run history, so you can see which procedures the team actually relies on.
Connectors
The connector permission model on Team plans is well designed and worth crediting. An Owner or Primary Owner enables a connector for the organization first, and enabling it grants nobody access because each person still authenticates individually. Then each tool gets a policy: always allow, needs approval, or blocked. Anthropic's line on that policy is the strong part. This applies org-wide to everyone using the connector, and individual users can't override it. Claude also inherits each person's permissions from the connected service, so nobody sees a document they could not already open. Type does not publish an equivalent per-tool organization policy.
Two sentences elsewhere in the docs undo it for teamwork. Connectors are only available in private projects. And: chats with synced content can't be shared. So the moment Claude reads from your Drive, your GitHub, or your Microsoft 365, the conversation lives in a private project and cannot be shared with anyone. Enterprise search inherits the same shape, and Anthropic notes conversations within your search project are private to you unless you manually share them.
That is the part most teams do not model in advance. The chats worth showing a colleague are exactly the ones that touched real company data, and those are the chats that cannot leave your account.
In Type, connections are assigned to a Space by an admin who connected the tool once for the organization. Personal connections stay separate from organization ones, each is revocable on its own, and reconnecting a broken one repairs it in place so every Space that used it keeps working. A thread that reads from Drive is a thread your Space can read, and an agent acts with the effective access available to the person and the conversation that started the work rather than everything wired into the workspace.
Slack
There is one place Claude is genuinely multiplayer, and Anthropic describes it well. In Claude Tag, everyone in a channel works with the same Claude, so anyone can steer it or pick up where it left off. That is the right idea. It replaced Claude in Slack on August 3, 2026, and on Team plans it is in beta. Two operational details: only a Primary Owner or Owner can set up its access and channels, and the Admin role can't; and channel usage is billed to the organization while direct messages are billed to the individual's own Claude account.
So collaborating on Claude Team means using a second product, in beta, configured by your most senior admin, on the one surface where it works. Your projects, your project memory, and your connectors do not follow you there.
Type connects to the Slack you already run and it is the same product either way. Mention an agent in a channel and it answers there with the same context, skills, and memory it has in Type. Route a Slack user group to a Space so the right people reach the right agent. DM Sidekick as a native Slack assistant. An agent mid-task can stop and ask a clarifying question through an interactive Slack modal, and the answer goes straight back into the running work.
The reason this matters is adoption. Teams do not fail with AI because the model is not good enough. They fail because the tool is one more place to check and the person who most needs it never opens it. Type does not ask your coworkers to move.
Cost
Run the numbers on ten people, using Anthropic's own worked example. Ten standard seats are $250 per month billed monthly, or $2,400 for the year billed annually. Premium seats put the same ten at $1,250 or $1,000. The floor is two members and the ceiling is 150 seats. Overage is opt-in: pre-purchased credits at standard API rates, with spend limits, and Owners can switch credits off entirely, in which case Anthropic notes members will be unable to continue working once they reach their limits and will need to wait for them to reset.
Give the plan its due on allowance. A standard seat includes 1.25x a Pro session's usage and a premium seat 6.25x, and because limits are per-member rather than pooled, nobody can eat your quota. For a team of ten heavy daily Claude users, that bundled allowance may well cost less per unit of AI than metering the same volume anywhere else, and we would rather say that than pretend otherwise.
Both products price access by team size, but each seat buys a different collaboration model. In Claude Team, another seat still gets a private chat rather than access to continue yours. In Type, every team member included in the plan can read and continue the same shared threads.
Type Basic is $50 per month for 2 team members and $100 of included AI usage. Pro is $100 for 4 members and $200 of usage. Enterprise plans cover 20, 40, or 60 members for $250, $500, or $750 with twice that price in included usage. AI is metered at the providers' published API rates, and requests through connected Claude or ChatGPT subscriptions do not use the workspace allowance.
Full comparison
Grouped by the decisions teams actually make when they evaluate the two.
Type
Claude Team
Who the work belongs to
A thread is the shared unit. Two people prompt into the same conversation, with a toggle for whether a message is for teammates or for the agent, so the discussion and the work live together.
Not available. Anthropic states chats within a project are not shared by default and that even in a public project your chats are private and inaccessible to other members unless you manually share them.
Nothing to share. Everyone with access to the Space already sees the session, including the agent's reasoning, the tools it called, and the files it produced.
You share a snapshot. Anthropic notes the chat snapshot includes all messages that were sent prior to sharing, so later turns stay private until you share again. Team and Enterprise users can only share with members of the same organization, not publicly.
Open the thread and keep going. The context, the connections, and the files are all still there, so a coworker finishes a half-done task without a handoff meeting.
A recipient of a shared chat can continue from it, and Anthropic documents that as starting their own copy. Their turns do not appear in yours.
Channels and a feed. Work is visible to the Space, so people notice what the agent is being used for and copy the pattern.
Owners and Primary Owners get usage analytics covering messages, active members, and adoption, and Anthropic notes Admins can't view them. That is aggregate usage rather than the work itself.
Memory and reusable knowledge
Memory is scoped to the workspace, the Space, or the person, so a correction one teammate makes carries into the next teammate's thread and into scheduled runs.
Anthropic states that Team plans do not have organization-level controls for memory features and that individual Team plan members manage their own memory settings directly. Each project has its own separate memory space.
One system for everyone in the workspace, generated in the background and read-only in the interface so nobody has to curate it by hand.
The older one. Anthropic says free, Pro, and Max users are being migrated to the improved memory experience and that in the interim, users on Team and Enterprise plans will stay on the legacy experience.
A Space carries its own files, connections, skills, memory, and automations, and everyone in the Space inherits all of it.
This is the plan's real sharing story and it works. Public projects are discoverable by the whole organization, private projects can be shared per person at can-view or can-edit, and sharing a project shares its knowledge base. Sharing a project with a group requires Enterprise and is in beta.
Skills have an owner and version history, a coworker's edit arrives as a change request the owner approves, and each skill reports 30-day usage, top users, and run history.
Skills and project instructions are real and travel with a shared project. Each person then runs them inside their own private chat, so use of them does not accumulate anywhere the team can see.
Where your team works
Mention an agent in a channel, route a Slack user group to a Space, or DM Sidekick as a native Slack assistant with the same context, skills, and memory it has in Type.
Claude Tag, in beta on Team plans, and it is genuinely multiplayer. Anthropic writes that everyone in a channel works with the same Claude, so anyone can steer it or pick up where it left off. Only a Primary Owner or Owner can configure its access and channels.
Full workspace in any browser, plus a Mac app where the agent can drive your own logged-in browser session for tools with no API.
Web, desktop, and mobile on every seat, which is broader coverage than Type publishes today.
The mobile browser works today and Slack covers phones. Type's iPhone app is still coming.
Native mobile apps are included with the seat. This one goes to Claude.
Every Space has a generated inbound email address, so someone with no account can start a thread. Schedules, webhooks, app events, RSS feeds, and Slack trigger work too.
A member starts a chat, or Claude Tag responds in a Slack channel. There is no inbound address or webhook a customer or a system can hit.
What the agent can do
Claude and Codex, with the harness, model, reasoning level, and speed selectable per thread, as a Space default, and per automation.
Claude, and Anthropic's own agent loop. If you want Anthropic's harness specifically, this is where to get it.
Type runs Claude and Codex for shared work, and the CLI can push a live local Claude Code or Codex session into a Type thread when a teammate needs to see it.
Anthropic states Claude Code is included with every Team plan seat, with premium seats offering more usage rather than more access. Terminal plus VS Code, Cursor, and JetBrains. At $20 a seat this is the strongest thing on the plan.
Agents work in a thread the Space can read, can pause to ask the channel a question, and sub-agents split large jobs and report back into the same thread.
Cowork covers this on Team plans and is capable. Anthropic's limitations list says Cowork sessions can't be shared with others, and the Cowork toggle is organization-wide.
Determined by the model you pick per thread, across Claude and Codex.
200k tokens on Team seats. Anthropic reserves the 500k window for Enterprise.
Six trigger types, including schedules, webhooks, app events, RSS, Slack, and inbound email, with each run posting into the channel where a teammate can read it or take it further. New automations are disabled by default.
Available through Cowork's scheduled tasks, which run with no device online. The run reports back into the private session of the person who created it.
Integrations and permissions
A broad catalog of native integrations across 16 categories, connected once and assigned to the Spaces that should have them, plus a custom API connector for anything else.
Anthropic's connector catalog plus MCP servers, with Google Drive, GitHub, and Microsoft 365 among the built-ins. Enterprise search spans them for Team and Enterprise plans.
An admin connects a tool once for the organization and assigns it per Space. Personal and organization connections stay separate, every connection is revocable, and a broken one reconnects in place.
Anthropic requires an Owner or Primary Owner to enable a connector for the organization before anyone can use it, and notes enabling it does not grant access because each person still authenticates individually. That is a sound design.
An agent acts with the effective access available to the person and the conversation that started the work, with file-level and folder-level scoping for sources like Google Drive.
Owners can set every connector tool to always allow, needs approval, or blocked, and Anthropic states this applies org-wide and individual users can't override it. Claude also inherits each person's permissions from the connected service. Type does not publish an equivalent per-tool org policy.
Connections are assigned to a Space, so the thread that uses them is visible to that Space by default. Connecting a tool does not make the work private.
The two collide. Anthropic states connectors are only available in private projects on Team and Enterprise plans, and that chats with synced content can't be shared. Enterprise search behaves the same way: conversations in your search project are private to you unless manually shared.
Administration
Spaces are the unit. Give a team a Space with the connections and skills it needs and leave the rest of the organization alone.
Seats are the unit, and some toggles are not. Anthropic states the Cowork toggle is organization-wide, so either all members have access or none do, and that Team plans lack the controls that would narrow it.
Members and admins are managed in workspace settings. Type does not ship SAML or SCIM today.
SSO and just-in-time provisioning are both available on Team plans, which is more than Type publishes. SCIM is not: Anthropic states it is not available for Team plans.
Workspace and Space membership with admin control over connections and settings.
Three fixed roles, owner, admin, and user. Anthropic's Team plan article lists role-based permissioning as a Team feature while its roles article heads that section custom roles, Enterprise plans only, so confirm before planning around it.
The thread is the record. Anyone with access to the Space reads what the agent did, in order, without asking for a log export.
Anthropic states audit logs are available for Enterprise organizations only, and the Compliance API is likewise Enterprise. On Team you have usage analytics, and Admins cannot see those.
Not applicable. Members are invited into the workspace.
Anthropic's pages disagree. The Team plan article lists Domain Capture as a Team feature; the domain article says claiming is supported on Enterprise plans only and that Team plans can verify a domain and block new personal accounts but admins can't claim or migrate existing accounts.
Cost
Basic is $50 per month for 2 team members and $100 of included AI usage. Pro is $100 for 4 members and $200 of usage. Enterprise plans cover 20, 40, or 60 members for $250, $500, or $750 with twice that price in included usage.
$25 per standard seat per month billed monthly, or $20 billed annually. Premium seats are $125 or $100. Anthropic requires a minimum of two members and supports up to 150 seats.
$250 per month on the Enterprise plan, which includes up to 20 team members and $500 of AI usage.
$250 per month billed monthly, or $2,400 for the year billed annually, which is Anthropic's own worked example. Ten premium seats are $1,250 or $1,000.
One pool the whole workspace shares, metered at the providers' published API rates with no Type markup, so an uneven week evens out across the team.
Per person, and this is a real advantage. Anthropic states usage limits on Team plans are per-member rather than applied to the team as a whole, and that one member hitting their limit does not affect anyone else's. Standard seats get 1.25x a Pro session and premium seats 6.25x, though the pricing page calls the same premium seat 5x a standard one.
Usage past the included allowance is metered at provider rates on the shared pool.
Opt-in pre-purchased credits at standard API rates, with spend limits, which Owners can disable entirely. Anthropic notes that with credits disabled, members will be unable to continue working once they reach their limits and will need to wait for them to reset.
Every person with workspace access counts toward the plan's included team members, whether they run agents every day or mostly read shared threads.
Everyone who needs access needs a seat. Anthropic also notes removing a member does not trigger an immediate refund because you have to reduce the seat allocation, and that it accepts cards only, not ACH.
Connect a personal Claude or ChatGPT subscription and requests that run through it are not metered against the workspace at all.
The seat is the subscription. A team member's personal Pro plan does not reduce what the organization pays.
This plan is built by the company that makes the model, and there is a lot on it we would not pretend to match. If your situation is one of these, buy the seats.
Three things to pin down with Anthropic before you sign, because their own pages disagree. The Team plan article lists Domain Capture as a Team feature while the domain article says claiming and migrating existing accounts is Enterprise only. The same article lists role-based permissioning while the roles article calls custom roles Enterprise only. And managing SSO appears under an Enterprise-only heading in one place while the SSO article says Team plans are supported. If any of the three matters to your rollout, get it in writing.
FAQ
Type runs Claude and Codex in channels your whole team shares, with memory that belongs to the workspace, integrations your admin assigns per Space, and the Slack you already use as a way in.
Get started for free14-day free trial. Plans start with 2 team members. Bring your own Claude or ChatGPT subscription.
