Claude Max and Team API Credits: Setup and Billing Explained

Claude Max and Team API credits help eligible subscribers pay for Claude inside their own apps. They are separate from the limits you see while chatting with Claude.

An API lets your software send a request to Claude and receive a response. For example, you could build a support tool that asks Claude to summarize a customer message.

This guide replaces our earlier Agent SDK credit-pool explanation. That older explanation should not be used to configure billing.

Claude subscription vs API: what is the difference?

Think about two tasks:

• You open Claude and ask it to rewrite an email.
• Your own website sends a customer’s message to Claude through an API.

The first is you using the Claude product. The second is your software using the Claude Platform. They have different usage and billing arrangements.

Anthropic’s release notes announced monthly API credits for Max and Team. You claim them by linking a Console organization through your plan’s billing settings. That link is the bridge between your subscription and your app’s API balance.

A Console organization is the account space where your team manages API access and billing. Before linking one, check its name and who manages it. This is especially useful if you have separate personal and work accounts.

How much credit is included, and what does it cover?

The following is a compact summary of Anthropic’s current credit rules:

Plan	Monthly credit
Max 5x	$100
Max 20x	$200
Team Standard	$20 per seat
Team Premium	$100 per seat

Team credits pool together, capped at $500. Free, Pro, and Enterprise are ineligible. New subscribers wait seven days.

Credits renew monthly, expire without rollover, and are used before purchased credits. They cover Platform API usage, Playground, Managed Agents, and Agent SDK usage. They do not cover interactive Claude Code, extra subscription usage, or third-party cloud platforms.

Self-run claude -p and Agent SDK runs use these credits with a linked organization’s API key. Subscription sign-in uses plan limits instead. GitHub Action, IDE, and desktop-launched runs are excluded from this benefit.

Without other funds, exhausted credits stop API requests. Purchased credits, auto-reload, or invoiced billing can continue paid usage.

Only one organization can be linked; changing it requires support. Max subscribers or Team Owners claim credits, with an Owner, Admin, or Billing role in Console. No payment card is required to claim them. The old June SDK credit is unavailable.

How to claim Claude Max or Team API credits

Use Claude in a web browser:

1. Open Settings > Billing for Max, or Organization settings > Billing for Team.
2. Find the API credits section and choose Link organization.
3. Select the intended Console organization, review the terms, and complete linking.
4. Check the balance in that organization’s Console billing page.

The official setup guide has the complete eligibility and linking details.

Before testing an app, write down which organization it should use. Then confirm that its API key belongs there. Do not paste the key into a chat to check it; inspect the key’s account or workspace in Console.

A simple budgeting example

Suppose you set a personal experiment budget of $12. Your first 20 test requests cost $0.30 in total.

The measured average is:

$0.30 / 20 requests = $0.015 per request
$12 / $0.015 = about 800 similar requests

These are invented numbers to show the calculation, not Claude pricing or a promise about how many requests you get.

Run a small sample of your real workload before estimating a month. Long inputs, larger answers, and repeated agent steps can change the cost. Recheck the estimate after changing the model or the task.

Keep an experiment budget separate from the credit balance. Having a balance does not mean you need to spend it all.

Does an API balance remove rate limits?

No. A balance tells you what funds are available. A rate limit controls how quickly you can send work.

Claude’s API has request and token limits. A request can hit a rate limit even when funds remain. Inspect the error and the rate-limit headers instead of assuming every failed call means an empty balance. For a rate-limit response, follow the retry guidance rather than repeatedly sending requests immediately. See the Claude API rate-limit documentation.

For a new app, I suggest starting with one test request, then a small batch. Confirm the response, the recorded usage, and the organization before enabling scheduled jobs.

A quick check when billing looks wrong

Ask these questions in order:

1. What ran? A chat, an interactive coding session, or your own API application?
2. How did it authenticate? Subscription sign-in or an API key?
3. Which organization owns the key? Check the organization you actually linked.
4. What does the error say? Save a redacted message and request ID; do not share the credential.
5. What is enabled in billing? Review available funds and any automatic purchasing settings.

This gives support a useful problem description. “My script failed” is hard to investigate; “this API request failed in this organization with this redacted error” is much clearer.

Where to go next

Read the Agent SDK cost tracking guide to measure your application’s usage. Before adding a key to a development project, read how to protect API keys when using coding agents. More implementation guides are in the Agent SDK articles.