Guides › New to AI

Claude Max API credits: what they are and what you can actually do with them

Updated Oct 22, 2026. The "how far it goes" figures are our own arithmetic from Anthropic's price list, and we say so each time.

Oct 30, 2026: Anthropic renamed Max 5x to Max and Max 20x to Ultra. Prices and API credits are unchanged. ^1,2

If you pay for Claude Max or Ultra, your plan now comes with US$100 or US$200 a month to spend on Claude's developer platform. It doesn't switch on by itself, it doesn't add to your normal Claude limits, and it disappears at the end of each month if you don't use it.

What the credits are

Anthropic sells Claude in two ways. There's the Claude app (claude.ai, the desktop and phone apps, and Claude Code), which you pay for with a monthly plan. And there's the Claude Platform, also called the API, which developers use to put Claude inside their own apps and tools. On the API you pay for exactly what you use, measured in "tokens" (small chunks of text).

Since October 7, 2026, Max and Team plans include a monthly allowance to spend on the API. Anthropic says the credits are "for running your own apps and agents on the Claude Platform." ⁴

^5,3,2 Anthropic states the amounts in US dollars.

The rules, in plain words

• Who gets them: the person who pays for the Max plan (on Team, the Primary Owner or an Owner). Your plan has to be active and in good standing, and new subscribers have to wait 7 days before they can claim. Plans bought in the iPhone or Android app count too, but you claim the credits on claude.ai in a web browser. ^3,5
• When they arrive: every billing cycle, shortly after your plan payment goes through. ³
• When they expire: at the end of each billing cycle. Unused credits don't roll over. ^3,5
• If you cancel: new credits stop when your plan ends. Credits you already have stay usable until they expire. ³
• If you upgrade from Max to Ultra: you get prorated credits right away, then US$200 each billing cycle. ³
• No card needed: you don't have to add a payment method on the Claude Platform to claim or use the credits. ³
• One place only: the credits go to a single Console organization. You can't split them, and everyone with an API key in that organization draws from the same balance. ³
• They can't be sold or given away: Anthropic's credit terms say credits have no cash value, are non-refundable and can't be transferred or sold. ⁷

What you can and can't spend them on

^3,5

How this differs from your normal Claude limits

The credits don't change your usage limits in Claude, Claude Code or Claude Cowork. ^2,3,5

Don't mix these two up: Claude also has "usage credits" (extra usage) that let paid plans keep chatting at API rates after hitting a limit. Your monthly API credits can't pay for that. ^2,8

How to claim them, step by step

Before you start: you need to be the Max subscriber, have been on Max for at least 7 days, and have the Owner, Admin or Billing role in a Claude Console organization. If you've never used the Console, you can create an organization during the claim. ^3,5

1claude.ai › Settings › Billing

→

2API credits › Link organization

→

3Pick or create a Console organization

→

4Accept the terms › credits appear in the Console

1. Open claude.ai in a web browser (not the phone app) and go to Settings › Billing. On a Team plan it's Organization settings › Billing.
2. In the API credits section, select Link organization.
3. Choose the Console organization that should get the credits, or create a new one. Choose carefully: you can link only one, and you can't change it yourself later. You'd have to contact support.
4. Review the Supplemental Credit Terms, then select Link organization.
5. Check it worked: sign in to the Claude Console at platform.claude.com and open the Billing page. Your credits show under Promotional credits with their amount and expiry date.
6. To use them outside the Playground, create an API key on the API keys page (platform.claude.com/settings/keys). Read API key safety first.

^3,5,9

Don't see API credits in your settings? Anthropic rolled them out over a few days, and the offer appears only after 7 days on an eligible plan. Anthropic suggests signing in to the Console with the same email you use on claude.ai, and trying again in a few hours if linking shows an error. If you're past 7 days on Max and still don't see it, contact Anthropic support. ³

5 things you can actually do with them

All five use only things Anthropic says the credits cover. Each has an honest difficulty rating and a rough idea of how far US$100 (Max) or US$200 (Ultra) goes.

How we work out "how far it goes": this is our own arithmetic from Anthropic's published API prices, not an Anthropic figure. Anthropic lists, per million tokens: Haiku 5.5 US$0.10 in / US$0.50 out (prompts up to 100,000 tokens; US$0.50 / US$2.50 above that), Sonnet 5.5 US$2 / US$10, and Opus 5.5 US$4 / US$20. Batch requests cost 50% less. Anthropic's own size examples: an average web page (10 kB) is about 2,500 tokens, and a research-paper PDF (500 kB) is about 125,000 tokens. Anthropic also notes that its newer models count about 30% more tokens for the same text, so treat every figure as a rough upper guide. ^10,2

1. Chat with Claude in the Console Playground (a backup for when you hit your limit) No code

What it is: the Playground is a web page in the Claude Console where you type a message, pick a Claude model and click Run. It's paid for by your API credits, which are separate from your plan's limits, so you can keep working there when the Claude app says you've hit your limit. It's also a quick way to try a different model on the same question. ^6,5

How to do it:

1. Sign in to the Claude Console and open the Playground (platform.claude.com/playground).
2. Type your message. You can add a "system prompt" with standing instructions, like "You are helping me edit a newsletter. Keep my voice."
3. Pick a model and click Run. The token count for each request is shown, so you can see what it cost.

Good to know: the Playground is plainer than the Claude app. It doesn't save your prompts or conversations on Anthropic's servers (your draft stays in your browser), so copy anything you want to keep. ⁶

How far it goes (our arithmetic): a request the size of one web page (about 2,500 tokens in) with a 500-token answer costs about US$0.01 on Sonnet 5.5, US$0.02 on Opus 5.5 and US$0.0005 on Haiku 5.5. That's roughly 10,000 such requests on Sonnet 5.5 for US$100, or 20,000 for US$200. Check the token count shown after each Run to see what you're really using.

2. Use Claude inside other apps that ask for an "Anthropic API key" A little setup

What it is: some chat apps, writing tools and other software let you plug in your own Anthropic API key so they can use Claude. Requests made with a key from your linked organization are Claude API usage, which the credits cover. ^3,5

How to do it:

1. In the Console, create a workspace for that app and give it a spend limit (see Spend limits).
2. Create an API key in that workspace, with an expiry date (see API key safety).
3. Paste the key into the app's settings, where it asks for an Anthropic or Claude API key. If it offers to "sign in with your Claude account" instead, that's a different route and may not use your API credits.

Be careful: Anthropic warns that giving your key to a third-party tool gives that tool's developer access to your Console account. Its advice: if you don't trust their reputation, don't trust them with your key. We don't recommend specific apps here, because we haven't checked each one. ¹¹

How far it goes: depends entirely on the app and the model it uses. Check the Usage page in the Console after a day of use to see your real cost. ¹¹

3. Power a free, open-source AI agent on your own computer A little setup

What it is: the free agents on our best free downloadable AI agents page cost nothing to download, but you have to bring your own AI model. All three can use an Anthropic API key:

• Hermes Agent (Nous Research): set ANTHROPIC_API_KEY and choose the Anthropic provider. Its docs say API-key requests are billed per token to that key's organization. ¹²
• OpenClaw: run openclaw onboard and choose "Anthropic API key". ¹³
• goose: in the desktop app, choose "Quick Setup with API Key" and paste your Anthropic key, or pick Anthropic in goose configure. ¹⁴

Important: pick the API key option, not "sign in with Claude", "Claude CLI" or "Claude Code login". Anthropic says that when you're signed in with your Claude plan, Agent SDK and claude -p usage draws from your plan's limits, not your API credits. Hermes's docs say its Claude sign-in route needs Max extra usage credits, which your API credits can't pay for. ^3,12

Be careful: these agents can run programs and change files on your computer. OpenClaw's own docs say its sandboxing is off by default. ¹⁵ Read each agent's security notes first, and put it in its own workspace with a spend limit. See our free agents page.

How far it goes: Anthropic's own worked example of a one-hour agent session on Opus 5.5 (50,000 input tokens, 15,000 output tokens) costs US$0.50 in tokens, or US$0.58 on Anthropic's Managed Agents, which adds US$0.08 an hour of runtime. At that size, US$100 covers roughly 170 to 200 sessions and US$200 roughly 340 to 400 (our arithmetic). Agents that read lots of files or web pages use far more tokens than that example. goose turns on Anthropic's prompt caching automatically, which cuts the cost of repeated context. ^10,14

4. Process a pile of documents at half price Technical

What it is: if you have a big, repetitive job, like summarizing 500 PDFs, tagging survey answers or pulling dates out of invoices, the Message Batches API lets you send them all at once and collect the results later, at 50% off the normal price. The credits cover it. ^16,10,3

How to do it: write the instructions once and test them on one document in the Playground. Then use the Playground's "code" button, which turns your request into a code snippet, and adapt it for the Batch API (or ask Claude to help write the script). Anthropic's batch processing docs have the details. ^6,16

No code? You can still do this one document at a time by pasting each into the Playground. It just costs full price, not half.

How far it goes (our arithmetic): a 125,000-token research paper with a 1,000-token summary costs about US$0.26 on Sonnet 5.5 at normal prices, or about US$0.13 in a batch. That's roughly 380 papers for US$100 at normal prices, or about 760 in batches (double that for US$200). For short items the size of a web page, Haiku 5.5 in a batch costs about US$0.00025 each, or roughly 400,000 for US$100.

5. Simple automations: run Claude on a schedule or from a script Technical

What it is: a small program that calls Claude for you, for example to sort incoming notes every morning or draft a weekly summary from a folder of files. The credits cover the Claude API, the Claude Agent SDK and Claude Managed Agents, and also claude -p (Claude Code's one-shot command mode) when you run it yourself with an API key from your linked Console organization. ^3,5

How to do it: start with Anthropic's Get started guide to make your first request, then use the Agent SDK or Managed Agents for anything with several steps. Use a separate workspace and key for each automation, and give each a spend limit. ^17,18,19,20

Watch out: runs started by the Claude Code GitHub Action, an IDE extension or the Claude desktop app count as Claude Code usage, so the credits don't cover them, even with -p. ³

How far it goes (our arithmetic): a daily job that reads about 10,000 tokens and writes 1,000 costs about US$0.03 on Sonnet 5.5 (about US$0.90 a month), or about US$0.0015 on Haiku 5.5. Small automations like that barely touch US$100.

What about Claude Code?

Does this cover Claude Code?

This is the most common confusion. Normal, interactive Claude Code isn't covered, in the terminal, code editor, desktop app or web. Anthropic says the credits are for building your own apps and agents on the Claude Platform, not for interactive Claude Code sessions or extra usage. ³

If the linked organization only has these credits and you try a Claude Code session with it, Claude Code shows "Credit balance too low". The exception is claude -p and the Agent SDK, run yourself with an API key from your linked Console organization. That counts as Agent SDK usage, which the credits cover. ^5,3

Other coding tools that take an API key: Anthropic's pages don't name third-party coding tools either way. A tool that sends requests to the Claude API with your key is Claude API usage. We couldn't find an Anthropic page confirming specific tools, though, so check the Console's Usage page after a short test.

Who should bother, and who can skip it

Claiming costs nothing and needs no card. Unclaimed or unused credits just expire each cycle. ³ Thinking about upgrading just for the credits? Compare Claude's plans first: Claude Pro vs Max.

Caveats: costs and spend limits

The API is metered. Every request has a price, and costs can keep going after your credits run out if you've added paid credits or turned on auto-reload. Here's how it works:

• With only the free monthly credits, you can't overspend. When they run out and the organization has no other credits, API requests stop until next month's credits arrive. Anthropic says usage is never charged to your Claude plan. ^3,5
• If you buy credits or turn on auto-reload, usage carries on and draws from those once the monthly credits are gone. ³
• The auto-reload catch: auto-reload is based only on your purchased balance. It can charge your card when that balance drops to your threshold, even while included credits remain. ⁵
• There's already a ceiling: organizations linked to an eligible plan move up to at least the "Start" tier, which has a US$500 monthly spend cap. ^5,21

How to set your own spend limits

1. Whole organization: in the Claude Console, go to the Billing page (platform.claude.com/settings/billing). In Spend limits, click Set limit (or Adjust limit) and enter an amount. When usage reaches it, requests stop until you raise the limit or the month resets. ²¹
2. Each app or agent: create a workspace for it on the Workspaces page (platform.claude.com/settings/workspaces). Select the workspace and click Spend limits to cap its monthly spending and set alerts. You can't set a spend limit on the Default Workspace, so make a new one. ^20,5
3. Keep an eye on it: the Console's Usage page (platform.claude.com/usage) and Logs page (platform.claude.com/logs) show what's being used. ¹¹

Timing differs: spend caps reset on the first of each calendar month (00:00 UTC), while your credits refresh on your billing date. ^5,21

API key safety

An API key is like a password that spends money. Anyone who has it can use your credits, or anything you've paid for on top. Anthropic's advice:

• Never share it, and don't paste it into public forums, emails or support tickets, even to Anthropic. ¹¹
• Only give it to tools you trust. Putting your key into a third-party tool gives that tool's developer access to your Console account. ¹¹
• Save it when you create it. The Console shows the full key (it starts with sk-ant-) only once. Store it somewhere safe, like a password manager. If you lose it, make a new one. ⁹
• Give each key an expiry date. The Console lets you pick 3 hours, 1 day, 7 days, 30 days or a custom time. ²²
• Use a separate key for each app or agent, ideally in its own workspace with a spend limit. Then one leak can't drain everything, and you can switch off one tool without breaking the others. ^11,5,20
• If you think a key has leaked, delete it right away: API keys page (platform.claude.com/settings/keys), the three dots next to the key, Delete API key. Then create a new one. ²³

Different rules apply: API use falls under Anthropic's Commercial Terms of Service, not the consumer terms of your Claude app. Those terms make you responsible for all activity on your account. ²⁴

Common questions

5 quick questions, answered

How much are the Claude Max API credits?

US$100 a month on Max and US$200 a month on Ultra. Team plans get US$20 per Standard seat and US$100 per Premium seat, pooled and capped at US$500 a month. Free, Pro and Enterprise plans don't get them.

Do Claude Max API credits roll over?

No. Unused credits expire at the end of each billing cycle and don't roll over. New credits arrive each billing cycle, shortly after your plan payment goes through.

Can I use the API credits in Claude Code?

Not for normal (interactive) Claude Code in the terminal, code editor, desktop app or web. Anthropic's help page says the credits do cover claude -p and the Claude Agent SDK when you run them yourself with an API key from your linked Console organization.

Will I be charged if I use up the credits?

Not on your Claude plan. Anthropic says usage is never charged to your Claude plan. If the Console organization has no other credits, API requests simply stop until next month's credits arrive. You only pay more if you buy API credits or turn on auto-reload.

Do the API credits give me more usage in the Claude app?

No. They don't change your usage limits in Claude, Claude Code or Claude Cowork, and they can't pay for extra usage in the Claude apps.

^3,5,2.

Related: Claude plans and prices · The same idea on ChatGPT Pro and Business · Get the most out of Claude · API prices for every model · Best free downloadable AI agents

New to AI?

AI crash course: start here

What AI chatbots are, how to ask, and how to stay safe. About 10 minutes.

Should I pay for AI?

What free plans really give you, what a paid plan adds, and who needs one.

Which AI model should I use?

Pick the right model and thinking level in 60 seconds.

Tip: subscribe on the website, not in the app

Buying through the iPhone or iPad app often costs more because of Apple's App Store fees. Sign up on the provider's own site (like claude.ai or chatgpt.com) and you can still use the app with the same account. How to switch and what to know · Compare app and website prices.