Claude API credits are money you’ve already paid for. If you’re on Claude Max or Team, your subscription now comes with credit for the Claude API every month. On Max 5x, it’s $100; on Team, up to $500 per organization. But the credit doesn’t show up in your account on its own, and whatever you don’t use expires at the end of the cycle.
Anthropic described the new feature in a help center article updated on October 7, 2026, and is rolling it out gradually, so some people don’t see it in their settings yet. I work with Claude every day, and one thing about the whole change stopped me short: you have to ask for the credit yourself.
How much API credit comes with your subscription?
Max 5x gets $100 a month, Max 20x gets $200, and Team gets $20 or $100 per seat. Pro, Enterprise, and the free plan get nothing. These amounts are current as of October 10, 2026, according to Anthropic’s documentation and pricing.
| Plan | Subscription per month | API credit per month |
|---|---|---|
| Max 5x | $100 | $100 |
| Max 20x | $200 | $200 |
| Team Standard | $25 per seat | $20 per seat |
| Team Premium | $125 per seat | $100 per seat |
On Team, the credit for all seats is pooled into one balance capped at $500. So a small business with two Standard seats gets $40. To reach the full $500, it would need 25 Standard seats or five Premium ones. I broke down the subscription prices themselves in a plan overview.
API credit
A prepaid amount for the Claude API, the access your own apps and automations use to call Claude instead of a chat window. It doesn’t behave like cash, though: under Anthropic’s terms, the credit can’t be refunded, transferred to anyone else, or sold.
You pay the full subscription whether you claim the credit or not. The only question is whether that money also gets you finished work.
What the credit covers and what it doesn’t
The credit covers the Claude API and the tools around it. It doesn’t cover chatting in the app or regular work in Claude Code.
According to the October 2026 documentation, the credit covers the Claude API including batch processing, Managed Agents, the Agent SDK, and the Playground, a sandbox for testing prompts in the Claude Console. It doesn’t cover interactive Claude Code in the terminal, an editor, the desktop app, or on the web. Claude Code run headless (with the claude -p command) using an API key from the linked organization is the exception.
Included credit gets used before purchased credit. Auto-reload, however, only watches the purchased balance, so it can kick in even while included credit is left. When the credit runs out and your organization has no other credit, the API stops until next month. Nothing gets charged to your subscription. Without purchased credit, spending can’t run away from you here the way it can with other items on your AI bill. The only risk is that the credit sits there unused.
How to claim the credit, step by step
Claiming it is a one-time link between your subscription and a single organization in the Claude Console. Check three conditions first. You need to have been on the plan for at least seven days. On Max, the subscriber links it; on Team, the Primary Owner or an Owner does. In the Console, you need the Owner, Admin, or Billing role. Sign in there with the same email you use on claude.ai. You don’t need to enter a payment card.
- On claude.ai, open Settings > Billing, or Organization settings > Billing on Team.
- In the API credits section, click Link organization.
- Pick the Console organization the credit should go to, or create a new one.
- Read through the Supplemental Credit Terms and confirm the link.
- You’ll then find the credit in the Console under Settings > Billing, in the Promotional credits section, along with the date it expires.
If you don’t see the option, you haven’t been on the plan for seven days yet, you’re missing the right role, or the feature hasn’t reached you yet.
Before you click Link organization
You only get one link. The subscription sends credit to a single Console organization, and you can’t switch it yourself; only Anthropic support can make that change. So link the organization where your business automations will run, not a test account you play around in.
What a small business can sensibly spend the credit on
Recurring text work that runs without a chat window and that someone does by hand today. It’s a good fit for sorting and summarizing incoming mail, drafting replies, translations, or pulling key points out of contracts and invoices, the way I describe in batch reading. For work that isn’t urgent, batch processing costs half as much. And before you put anything into production, test the prompt in the Playground.
The credit should also cover an AI step in n8n or Make, as long as it calls the Claude API with a key from the linked organization. The documentation doesn’t name these tools explicitly, though, so check it on one automation first.
Workspace
A separate part of an organization in the Claude Console with its own API keys. All workspaces draw from the same credit, but you can set a spending limit for each one. If one automation gets stuck in a loop, it won’t drain the credit for everything else.
How much work fits into $100?
By my estimate, $100 covers roughly eleven thousand email summaries if you run them through Sonnet 5.5. I’m using pricing as of October 10, 2026, and counting generously on purpose: a 500-word email plus the instruction comes to about two thousand input tokens. A 150-word summary comes to five hundred output tokens. A token is a piece of a word, and the API is billed by tokens.
| Model | Price per million tokens (input / output) | Summaries for $100 |
|---|---|---|
| Haiku 5.5 | $0.10 / $0.50 for shorter requests | about 222,000 |
| Sonnet 5.5 | $2 / $10 | about 11,000 |
| Opus 5.5 | $4 / $20 | about 5,500 |
A business that gets a hundred emails a day receives about three thousand a month. Summarizing them with Sonnet would take roughly $27 out of the Max 5x credit. So sorting the mail, a job someone would otherwise spend every morning on, can be covered by credit you’ve already paid for.
Who won’t get anything out of the credit
The credit doesn’t apply to the Pro plan, and switching to Max for its sake makes no sense. On Max, the credit equals the subscription price, so you could buy the same amount of API usage on its own.
It also does nothing for a business that has no work for the API yet. Building an automation just to spend the credit is worse than letting it expire. Look at it the way you would in a subscription audit: what did it bring you last month?
There’s also a trap for anyone who works for clients, like a marketing agency. Under the terms, the credit can’t be transferred to anyone else, so you can’t give it to a client for their account.
The whole thing in a nutshell
- Only Max and Team get the credit, up to $500 a month per organization.
- It covers the Claude API, agents, and the Playground, not chat or regular Claude Code.
- You claim it after seven days on the plan, under Settings > Billing > API credits.
- Choose the Console organization carefully; you can’t switch it yourself afterward.
- Give each automation its own workspace with a spending limit.
- Unused credit expires, so put it to work on recurring text tasks.
A $100 monthly credit that would otherwise quietly expire covers thousands of email summaries. If you’d like help choosing that first job, we can go through it together in a no-obligation consultation. Then I can teach your team to work with Claude through Claude training for teams. All it takes is one link in your settings and one job nobody has to do by hand anymore.
