Where you see and cap Claude Code spend depends on how your team signs in. Teams and Enterprise plans use the claude.ai admin console and per-seat allowances. Console API users use workspace spend limits. Cloud provider users rely on their cloud billing console, with OpenTelemetry or a gateway for per-user numbers.
The three setups
Per Anthropic's cost documentation:
| Setup | See spend | Cap spend |
|---|---|---|
| Teams or Enterprise | Spend report in org analytics | Spend limits in admin settings |
| Claude Console (API) | Console usage page | Workspace spend limits |
| Bedrock, Vertex or Foundry style cloud | Your cloud billing console | Your cloud budget controls |
On Teams and Enterprise, usage draws from a per-seat allowance that resets on a rolling five hour window and a weekly window, shared with Claude chat. To let members continue past it, turn on usage credits and set limits at organization, group or member level.
Sizing a pilot
The page states that across enterprise deployments the average is around 13 dollars per developer per active day, and 150 to 250 dollars per developer per month, with costs below 30 dollars per active day for 90 percent of users. It then advises starting with a small pilot group to set a baseline rather than extrapolating from the average. Treat those figures as Anthropic's description of its own deployments, not as a forecast for your team.
Rate limits on the API
For Console organizations the docs give tokens per minute and requests per minute starting points per user, falling as team size grows because fewer people use it at once. For 20 to 50 users they suggest 50k to 75k tokens per minute per user. They apply at the organization level, so one person can briefly exceed their share. You can set a lower workspace rate limit on the automatically created Claude Code workspace to protect other production workloads.
Per-user numbers
- Teams: export the spend report CSV.
- Enterprise: the Enterprise Analytics API returns per-user usage and cost.
- Console: the Claude Code dashboard and the Claude Code Analytics API.
- Any setup: OpenTelemetry streams per-user metrics into your own stack in near real time.
Reporting contracted rates
By default, the figures Claude Code shows use list price. If you pay contracted rates, an admin can set the modelPricing managed setting, and the cost line then notes your configured rates. It changes what is displayed, not what is billed, and the figure remains an estimate.
Common causes of surprise bills
The docs trace unexpectedly high spend on API or cloud plans to long sessions that were never cleared, or to the most expensive model left as default. Both are habits, and a short onboarding note fixes most of it. The next post covers them.
Quick check
Is the cost shown in the session an invoice?
No. It is a local estimate at list price unless a pricing table is configured. For billing, use the Console usage page or your admin report.
How do I get per-user cost on a cloud provider?
Use OpenTelemetry export, a self-hosted gateway, or an LLM gateway that tracks spend per key.
Sources
Last reviewed: 19 October 2026.
