Claude Code Enterprise Pricing and the Security Controls to Budget Beside It
Try it live
Watch AgentShield block an attack in real time.
Pick a scenario and drive the inspection lane yourself. No signup needed.
Run a request
Runs the live engine on your text. Nothing is stored, no account needed.
Inspection lane
INSPECTINGPolicy trace
High-risk action held for approval
Audit trail
- § · → → →
Claude Code is included in every paid Anthropic plan, so the seat price is the smaller part of what an enterprise deployment costs. On Anthropic's pricing page, read on October 30, 2026, the Enterprise plan is listed at US$20 per seat per month billed annually with a 20 seat minimum, and usage is billed at API rates. The larger lines are usage, the engineering time to run the controls, and whatever you add to cover actions the controls do not judge. Check the page again before you sign, because plan terms change.
What Team and Enterprise include for Claude Code
Anthropic states that Claude Code is included in all paid plans and shares the same usage limits as the rest of the plan, so chats and Claude Code draw from one pool. The table below summarizes the two business plans from the pricing page.
| Item | Team | Enterprise |
|---|---|---|
| Seat price | Standard seats $20 annual or $25 monthly. Premium seats $100 annual or $125 monthly | US$20 per seat per month, billed annually, 20 seat minimum |
| Claude Code | Included | Included |
| Usage | Seat limits, with usage credits available. Up to $500 per month in pooled API credits is listed | Billed at API rates, scaling with the model and the task |
| Admin and security | SSO, central billing, usage analytics, connector admin controls | Adds role based access, SCIM, audit logs, a Compliance API, custom data retention, network level access control and IP allowlisting |
| Spend controls | Listed as available | User and organization spend limits |
Two details matter for budgeting. Team does not list audit logs, the Compliance API or custom data retention, so a security review that asks for them points to Enterprise. And Enterprise usage is variable, so finance needs a ceiling, not a seat count.
The costs that sit outside the seat price
None of these lines appear on Anthropic's pricing page, and all of them show up in the first year.
| Line | Who pays it | What to estimate |
|---|---|---|
| API rate usage | Finance | Heavy agentic sessions cost far more than chat. Pilot with a cap, then set the organization spend limit from real numbers |
| Managed settings and policy upkeep | Platform engineering | Someone owns the managed file, the MCP allow list and the model lists, and updates them as engineers ask for exceptions |
| Sandbox and network rules | Platform and security | Allowed domains for package registries and internal hosts take trial and error to get right |
| Telemetry pipeline | Security operations | OpenTelemetry metrics need a destination, dashboards and a person who reads them |
| A control for production actions | Security | Managed settings govern what runs on a laptop. Deploys, payments and access grants need a per call decision |
What Anthropic's controls cover, and where the last line begins
Anthropic's documentation describes managed settings that developers cannot override, allow, ask and deny permission rules, a sandboxed bash tool, managed MCP configuration and model allow lists. It also says plainly that "a Bash command you approve can still write anywhere your user account can" and that no system is completely immune to attacks. That is an honest description of a client side control. The page that walks through each control and its gap is our Claude Code enterprise security guide, and it ends with a table of situations where you should buy nothing more.
The last budget line is for teams whose agents act on systems outside the laptop. If Claude Code, or an agent you build with the Agent SDK, can call a deploy API, a ticket system or a payment tool, a prompt injection in a README or a web page can steer an approved call. A gateway that checks the arguments of each call, holds high impact ones for a named approver and writes one attributed record fits there. Our published plans are on the AgentShield pricing page, so you can put a real number next to the engineering time it replaces.
What moves the bill after rollout
Three things change the number more than the seat price does. The first is model choice. Anthropic bills Enterprise usage at API rates that scale with the model and the task, so a team that defaults every session to the largest model pays for it on every long run. Anthropic's managed settings include availableModels and deniedModels, which let an administrator set which models developers can pick, and that is as much a cost lever as a security one.
The second is autonomy. A developer who approves each step stays in the loop and keeps sessions short. A team that moves to auto mode, or runs the agent headless in a pipeline, lets one task run for far longer and spend accordingly. That is the right trade for many jobs, and it is also the point where the interactive prompts that protected the session are gone. Budget for a decision on each action at that stage, not only for the tokens.
The third is incident cost. A single bad action on a production system, such as a deploy to the wrong environment or a bulk edit of customer records, costs engineering days to unwind and may trigger a customer notice. Seat prices are small next to that, which is why security teams compare a per call approval control against the cost of one such event rather than against the seat price.
A simple way to size the first year
- Multiply seats by the plan price. Use Enterprise if an auditor will ask for audit logs.
- Run a four week pilot with fifteen to thirty engineers and read the actual usage at API rates. Multiply by the roll out headcount, then add a margin for heavy users.
- Set the organization spend limit slightly above that figure, and review it monthly. Tracking this next to your other cloud and SaaS bills in a read only view of cloud and SaaS spend keeps the number in front of finance instead of in a single admin console.
- Estimate platform engineering time for managed settings, MCP review and sandbox rules. A week to set up and a few hours a month is a realistic starting guess for a team of a couple hundred engineers, but measure yours.
- List the production actions an agent may take. If the list is not empty, price a per call control for those paths.
When the seat price is the whole story
If engineers use Claude Code on application code, the work stays on laptops and in repositories, and no agent holds credentials for production systems, then the seat price, a usage cap and Anthropic's managed settings are a complete answer. You do not need to buy anything from us. The budget grows when agents gain write access to deploys, tickets, billing or customer data, or when they run headless in pipelines where Claude Code's interactive prompts never appear. For that case, see human approval for AI agents for how to decide which actions deserve a person, and our coding agents page for the wider control set.
Is Claude Code included in the Enterprise plan?
Yes. Anthropic lists Claude Code as included in both the self serve and sales assisted Enterprise tiers, and in Team. It draws from the same usage pool as chat, and Enterprise usage is billed at API rates.
How much does Claude Code cost per seat for a company?
Team standard seats are listed at $20 per seat per month billed annually, or $25 monthly, with premium seats at $100 and $125. Enterprise is listed at US$20 per seat per month billed annually with a 20 seat minimum, plus usage at API rates.
See the firewall block an attack live.
Drive the Threat Console and watch a real prompt injection get stopped, then put AgentShield in front of your own agents.