If you pay for Claude Max, Ultra or Team, you have monthly credit for the Anthropic API that you are probably not using. Claim it today: linking takes two minutes, it covers the API and the Agent SDK, it does not cover interactive Claude Code, and whatever you do not spend in the month is gone.
What it is
Anthropic gives credit on the Claude Platform, its developer platform with the API and the Console, to people on certain Claude plans. According to the official documentation, the amounts are:
| Plan | Monthly credit |
|---|---|
| Max | $100 |
| Ultra | $200 |
| Team, Standard seat | $20 per seat |
| Team, Premium seat | $100 per seat |
On Team, credit from every seat is pooled into one balance capped at $500 a month. The docs’ own example: three Standard seats and two Premium seats get $260 a month. Discounted Team plans (nonprofits, scientists) get the same.
Free, Pro and Enterprise are not eligible. If you just subscribed, you have to wait 7 days on the plan before claiming.
What matters is what it covers:
| Product | Covered? |
|---|---|
| Claude API | Yes |
| Claude Managed Agents | Yes |
| Claude Agent SDK | Yes |
| Console Playground | Yes |
| Claude Code | No |
| Extra usage in the Claude apps | No |
| Claude on AWS Bedrock, Google Cloud or Microsoft Foundry | No |
The Claude Code part is what trips people up. The support article puts it this way: claude -p (non-interactive mode) and the Agent SDK spend this credit only when run with an API key from the linked organization. Signed in with your plan, you use plan limits as usual. Runs from the GitHub Action, the IDE or the desktop app count as Claude Code usage, so they are not covered either.
How to claim it
You need to be the plan subscriber (Max or Ultra) or an Owner of the Team plan, and hold the Owner, Admin or Billing role in a Console organization. If you have no organization, create one at platform.claude.com or during the claim flow.
- On claude.ai, go to Settings > Billing (Max or Ultra) or Organization settings > Billing (Team).
- In the API credits section, choose the Console organization that will receive the credit and accept the terms.
- Check that the balance shows up under Promotional credits on the Console billing page.
- Create an API key in that organization and make your first call.
Careful with step 2: each plan links to one organization and you cannot change it yourself. Changing it means contacting support. Choose before you click.
For a script to use the credit, all it needs is that API key:
export ANTHROPIC_API_KEY="sk-ant-..." # key from the linked organization
claude -p "summarize the errors in logs/yesterday.log in five lines"
Without the variable, claude -p would use your plan session and leave the credit alone.
Four details worth knowing
- It expires every month. Credit arrives each billing cycle, shortly after payment is processed, and anything unused does not roll over. On annual plans it still arrives monthly.
- It is spent first. If you also bought credit, the included credit goes first and yours after.
- The balance is shared. Every API key and workspace in the organization draws from the same balance. To cap what one project spends, give it its own workspace with a spend limit.
- When it runs out, it stops. With no purchased credit or auto-reload, API calls fail until next month’s credit arrives. Nothing is ever charged to your Claude plan, so no surprises on your card.
Linked organizations also move up to at least the Start usage tier, and to the Build tier when monthly credit exceeds $200. That affects rate limits, not the balance.
OpenAI has just launched something similar for its $100, $200 and $500 plans. Simon Willison sums it up as “pretty much exactly the same deal”.
My take
If you build on your own, this is the first thing I would switch on this month. You already pay for the plan, and $100 a month of API goes a long way: the scripts that run overnight, an Agent SDK agent that sorts support email, or the demo you never published because you did not want a separate API bill.
I would use it for three specific things:
- Agent SDK automations you do not have today because adding another card felt like a chore.
- Client prototypes before the client sets up their own account. You test the real cost on their case before quoting.
- Batch jobs. The Message Batches API costs 50% less according to the docs, and it is the cheapest way to stretch the credit.
What I would not do is rely on this credit for production. It expires, it depends on staying on the plan, and if it runs out mid-month the API stops. For a client, that means their automation goes silent on the 23rd. For production, use a separate organization with its own billing and alerts. The credit is for building and testing.
And split workspaces from day one. If a script gets stuck in a loop, I would rather it burn through its own workspace limit and not the balance everyone else works from.
Sources
- Claude Platform: API credits for Max and Team plans
- Claude Help Center: Monthly API credits for Max and Team plans
- Simon Willison: API credits benefit for eligible ChatGPT subscriptions
If you want to see how I build agent automations without a team and without blowing the budget, follow me on LinkedIn.