How to cap your Claude API spend

How Anthropic Usage Tiers Cap Your Spend

The Claude API enforces two distinct kinds of boundaries: rate limits, which govern the number of requests you can fire over a short window, and spend limits, which dictate the maximum bill you can run up in a calendar month. Every organization belongs to a usage tier. Anthropic assigns this tier automatically based on your account standing and usage history, and you move up as you use the API over time. If you just opened an account, you might start in the Evaluation tier, which has limits set below the standard tiers while Anthropic establishes your account history.

Standard organizations sit on one of three fixed tiers, or a custom tier negotiated directly with Anthropic.

You can check your current tier and active limits anytime at platform.claude.com/usage/limits. If you need more room on a standard account, click "Request tier increase" directly on that page or contact Anthropic support.

Things work slightly differently if you use Claude Platform on AWS. In that setup, you pay through AWS Marketplace invoices rather than buying Anthropic credits directly, your account starts on the Start tier, and you move up automatically as you pay your AWS Marketplace bills. Per-workspace rate limits, fast mode, and the in-console tier request flow are not available on Claude Platform on AWS; if you need higher limits there, you have to contact support or your account rep directly.

Setting a Custom Spend Limit for Your Organization

The default tier caps are high enough that a runaway script can still cost you hundreds of dollars before Anthropic steps in. If you are on the Build tier with a $1,000 limit but your monthly side-project budget is $150, you should lower the organization ceiling yourself. A custom spend limit cannot exceed your current tier's cap, but you can set it to any number below that cap.

Steps to adjust the organization spend limit

1. Sign in to the Claude Console and navigate to the Billing page at platform.claude.com/settings/billing.
2. Find the Spend limits section on the page.
3. Click "Adjust limit" (or "Set limit" if you have never configured one before).
4. Enter your target monthly spend limit in USD and save the change.

Once saved, this number acts as your hard monthly cap. If you ever need to restore access early after running into this ceiling, you can raise or remove the limit right from this screen.

Setting Per-Workspace Spend Limits

Workspaces let you isolate API keys, team members, and limits by environment or project. Only Organization Admins can create new workspaces, and accounts are limited to 100 workspaces by default, though archived ones do not count against this total.

Every organization comes with a Default Workspace. You cannot rename the Default Workspace, change its color, archive it, or delete it, and you cannot set spend limits on it. For any other workspace you create, you can define an isolated monthly spend limit and set up threshold email alerts.

Configuring workspace limits and alerts

To set limits for a specific workspace, head to platform.claude.com/settings/workspaces and click on the target workspace:

• In the side panel that opens, click "Spend limits".
• Click "Change limit" under Monthly limit to set a cap in USD. The workspace limit must be lower than your organization limit. If left blank, it simply inherits the organization's limit.
• Click "+ Add notification" under Email notification to trigger an email whenever workspace spend reaches a specific threshold.

Workspace limits are strict boundaries for the keys inside that workspace, but organization-wide limits always apply regardless of how you distribute them. If you have three workspaces with $400 limits on an organization capped at $500, the entire account halts the moment total spend across all three reaches $500.

To see which workspace paid for a specific call in your logs, look at the anthropic-workspace-id response header returned with each request. You can also view workspace usage and costs directly on the Usage page at platform.claude.com/usage or the Cost page at platform.claude.com/cost. Anyone with the Organization Billing role can see cost, usage, and limit values across all workspaces.

If you use Claude Code, Anthropic automatically provisions a dedicated Claude Code workspace when an organization member first signs in. This workspace handles limits on its own, and it is the only workspace that supports per-user monthly spend limits. If you ever need to shut down an obsolete workspace permanently, archiving it deactivates the workspace and invalidates every API key inside it within seconds, though historical reporting data is kept.

What Happens When You Hit a Limit

When your account spends through a limit, requests do not silently queue or bill in arrears. They fail immediately. The exact error code and headers depend on whether you hit an Anthropic tier cap, a custom spend limit, or a Claude Code workspace limit.

Tier caps vs custom spend limit errors

When you hit a tier's monthly spend cap:

• The API pauses requests until 00:00 UTC on the first day of the next calendar month.
• Requests return an HTTP 429 response with an error type of rate_limit_error.
• There is no retry-after header in this 429 response. Standard SDK retries will fail continuously until access resumes or you move to a higher tier.
• On the Messages API, the response body includes error.details.error_code: "enforced_spend_limit_reached". Check this string in your error handling so your app does not treat the failure as a temporary traffic spike.

When you hit a user-configured spend limit (organization or workspace):

• Requests return an HTTP 400 response with an error type of invalid_request_error.
• The error message text explicitly identifies what tripped. For organization limits, it begins with: You have reached your specified API usage limits.
• For workspace limits, it begins with: You have reached your specified workspace API usage limits. Both messages indicate when access resumes.

Requests over limit in the Claude Code workspace are checked separately and can return an HTTP 429 error that includes a retry-after header.

Tracking Spend Programmatically

If you prefer to monitor workspace costs via internal tooling or automated pipelines instead of clicking through the console, you can use Anthropic's Usage and Cost Analytics API. The endpoint lets you slice historical spend by workspace over custom time buckets.

Here is a query that pulls workspace usage using the analytics endpoint:

curl --globoff "https://api.anthropic.com/v1/organizations/analytics/usage?\
starting_at=2026-09-01T00:00:00Z&\
ending_at=2026-09-08T00:00:00Z&\
workspace_ids[]=wrkspc_01JwQvzr7rXLA5AGx3HKfFUJ&\
group_by[]=workspace_id&\
time_bucket_width=1d" \
  -H "anthropic-version: 2026-10-21" \
  -H "anthropic-beta: analytics-usage-cost-2026-08-24"

You can also read your current organization and workspace rate limits or update workspace rate limits programmatically using the Rate Limits API.

Monitoring Claude API Spend with Costbrake

Native console limits protect your wallet when things go completely off the rails, but they only trip after the money has already left your account. If you want early warnings before a billing threshold hits, Costbrake watches what you spend across 23 cloud and AI providers, including Anthropic, AWS, OpenAI, and OpenRouter.

Costbrake runs checks hourly on its Free plan and every 5 minutes on Pro and Team, checking up to every minute during an active spike. It fires spike alerts when an hour costs more than 3x the usual 7-day baseline and at least $5 more than average. It also tracks default budget thresholds at 50%, 80%, and 100%, and warns you if the pace of the last 72 hours projects that your month-to-date spend will exceed your budget before month-end.

For Anthropic, Costbrake connects via an Admin key created in the Claude Console under Settings > Admin keys. Because Anthropic does not provide read-only keys, Costbrake uses this key to pull usage and cost report APIs, and pairs with an opt-in killswitch feature called brakes. If an API key runs rogue, you can trigger a brake from the dashboard to disable that single key or disable all active keys in an entire workspace (up to 60 keys per brake) without logging into the Claude Console. Brakes are fully manual and require admin confirmation; Costbrake never disables keys on its own.

Setup takes a couple of minutes: you can follow the agent setup flow to let your coding agent handle configuration, or connect keys directly at app.costbrake.com/signin. You can compare plan tiers on our Costbrake pricing page.