Claude Max and Team API Credits: How to Claim and What They Cover

Anthropic now includes a monthly pool of Claude API credits (first noted in our Haiku 5.5 launch coverage; this is the claim-and-limits guide) with its Max and Team plans. Max 5x subscribers get $100 a month, Max 20x subscribers get $200, and Team plans get $20 per Standard seat or $100 per Premium seat, pooled into one balance capped at $500 a month. The terms are on Anthropic's API credits for Max and Team plans documentation page and in the help center article.

The headline is easy to misread. Some summaries say "$500 in monthly API credits for Max and Team." That figure is the ceiling on the Team pool, not what a Max subscriber receives. And the credits cannot pay for interactive Claude Code, which is what most Max subscribers use most. This guide covers who gets what, how to claim, the exclusions, and how to avoid wasting a credit that expires every month.

TL;DR: the questions people are asking

Who gets how much

Anthropic's documentation gives the amounts in a single table.

On Team plans the credits for all seats go into one pooled balance, and the pool is capped at $500 per month. The first credit and each later one are based on the seats you have at the start of that billing month. Anthropic's worked example: three Standard seats and two Premium seats gives $260 a month, and adding one more Premium seat raises the next month to $360.

Do the math on your own team. Five Premium seats would be $500 and hit the cap exactly, so a sixth Premium seat adds nothing to the credit. If you have many Standard seats, you reach the cap at 25 of them. Seats you remove stay on the plan until renewal, so the credit may fall after that.

What the credits cover, and what they do not

This is the part that decides whether the credits are useful to you.

The pattern is clear. The credits are for building your own applications and agents, not for using Anthropic's own tools. Anthropic's docs say you "use them to build and run your own applications and agents." If your Max plan is mostly Claude Code usage, the credits do nothing for that usage, and they do not raise your Claude or Claude Code usage limits either.

One edge case from the help center: claude -p (print mode) is covered only when you run it yourself with an API key from your linked Console organization. If you are signed in with your Claude plan, it uses plan limits instead. Runs from the Claude Code GitHub Action, IDE extensions or the desktop app count as Claude Code and are not covered, even with -p.

How to claim them, step by step

You need two roles: the right role on the Claude plan, and a role on a Claude Console organization.

Then:

1. On claude.ai, go to Settings, then Billing (Max) or Organization settings, then Billing (Team).
2. In the API credits section, choose Link organization.
3. Pick an existing Console organization or create one during the flow.
4. Accept the supplemental credit terms and confirm.
5. Find the credit in the Console under Settings, then Billing, in the Promotional credits section, with amount and expiry date. Create an API key and make a request.

Three details cause most of the support tickets.

• Seven-day wait. New subscribers can claim after seven days on an eligible plan.
• One link, forever. Each plan links to one Console organization, and each organization can receive credits from only one plan. You cannot change the link yourself; you must contact support. Link the organization where you will actually build.
• Claim in a browser. Plans bought through the iOS or Android apps are eligible, but you must claim on claude.ai in a web browser.

Anthropic also says rollout takes several days, so if you are eligible and do not see the section yet, check your plan, role and subscription age before contacting support. Use the same email for the Console as for claude.ai.

How the credits behave once you have them

• Timing. Credits arrive each billing cycle, shortly after the plan payment is processed. Annual plans get them monthly.
• Expiry. Unused credits expire at the end of the cycle and do not roll over.
• Order. Included credits are spent first, then any credits you bought. Auto-reload looks only at your purchased balance, so it can reload even while included credits remain.
• Sharing. Every API key and workspace in the linked organization draws from the same balance.
• Spend caps. Credit-funded usage counts toward your organization's monthly spend cap, which resets on the first of the calendar month while credits refresh on your billing cycle. The two calendars do not line up.
• Tier bump. Linked organizations move up to at least the Start tier, and to at least the Build tier if monthly credits exceed $200.

When the balance runs out and there are no purchased credits, API requests stop until the next grant. Anthropic is explicit that usage "is never charged to your Claude plan." The error you will see is Your credit balance is too low to access the Anthropic API. Please go to Plans & Billing to upgrade or purchase credits. If you try a Claude Code session against an organization that only holds these credits, Claude Code shows Credit balance too low · Add funds.

Make a monthly $100 or $200 go further

Because the credit expires, the goal is to use it on work that benefits from the API, not to bank it. Anthropic's own suggestions are sound.

1. Pick the model per task. Use a smaller model for classification, extraction and routing, and keep the largest model for the hard steps. See our Sonnet 5 permanent pricing breakdown for the current price ladder.
2. Use prompt caching for repeated context such as long system prompts and shared documents.
3. Use the Message Batches API for work that can wait. It costs 50 percent less.
4. Give each project its own workspace and set a spend limit, since all keys share one balance.

A rough sizing: $100 buys a few million output tokens of a mid-tier model, or far more of a small one, so a side project, an internal tool or an agent prototype fits easily. An always-on production agent will not.

Why this matters now

Two things changed around the same time. First, plan users have complained about opaque limits and charges, as in our coverage of repeated charges after disabling usage credits and the Claude Code limits change. Second, Anthropic has been pushing builders toward the Agent SDK and Managed Agents. A monthly credit that can only be spent on those surfaces nudges subscribers from using Claude to building on it.

The FAQ in Anthropic's help center also notes a related point: an earlier Agent SDK credit from June is not available, and these API credits are a separate offer. If you remember a June credit tied to claude -p, do not expect it to continue under the same terms.

What we could not confirm

• Whether the credit amounts will change. Anthropic says to check the help center before relying on figures.
• Any end date for the program. The documentation we read states none.
• How the credits interact with enterprise invoicing beyond the line that invoiced organizations are billed normally above the credit.

What this means for what you build or pay

• Max subscribers: link a Console organization this week, then decide which side project or internal agent uses the credit. A credit you never claim is $1,200 to $2,400 a year of unused value.
• Team owners: count seats against the $500 cap before adding Premium seats to raise the pool.
• Claude Code heavy users: this does not change your plan limits. Do not cancel an API budget thinking the credit replaces it for Claude Code.
• Teams on Bedrock or Vertex: the credit does not apply there. Only the Claude API in the Console qualifies.

Related reading on explainx.ai

• Claude Haiku 5.5 launch, pricing and benchmarks
• Claude users reporting repeated charges after disabling usage credits
• Claude Code limits and the September change
• Sonnet 5 permanent pricing
• Claude Managed Agents session viewer and auto mode
• Build useful AI agents with Claude Code
• AI credit resale market and token safety

Amounts and terms are from Anthropic's documentation as read on October 8, 2026 and may change; check the help center before relying on them.