Maxing out your Claude Max

Most people pay for Claude Max for one reason: more usage. That is real, but it is the smallest part of what the plan includes. The rest sits in settings nobody opens, including a monthly credit worth as much as the plan itself that is easy to miss.

Here is everything in the box, how to claim the part most people miss, and what a month of heavy use taught us about spending it well. Plans change; this is how they stand on 23 October 2026, from Anthropic’s own pages.

What you pay for

Max comes in two sizes: Max 5x at $100 a month and Max 20x at $200 a month, with five or twenty times the Pro plan’s usage per session.¹ Usage runs in five-hour sessions that reset on their own, plus a weekly limit across all models.¹

Beyond the volume, both tiers include:¹ ²

• Priority access at busy times, so your work keeps moving when demand peaks.
• Early access to new models, features and products.
• Higher output limits, for long documents and large pieces of code in one go.
• Every model: Opus, Sonnet and Haiku, plus Fable, which can use up to half of the weekly limit. Context windows reach up to a million tokens, depending on the model.

Claude Code is in the same subscription

The plan includes Claude Code, Anthropic’s coding agent for the terminal and the editor, with no separate bill.¹ It draws on your plan’s usage limits.¹

For a small team this is the biggest line on the invoice. Claude Code reads the project, makes the change, runs the tests and explains what it did. Our own site, its audits and its admin are built this way, with an engineer deciding and checking every change.

The perk most people never claim: API credits

Max now includes monthly credits for the Claude API: $100 a month on Max 5x and $200 a month on Max 20x.³ They pay for your own applications and agents, not for chat.

What they cover: the Claude API, Claude Managed Agents, the Claude Agent SDK and the Playground in the Claude Console.³ What they don’t: Claude Code (it stays on your plan’s limits) and extra usage in the Claude apps.³

How to claim them:³

1. Wait until your plan has been active for 7 days.
2. On claude.ai, in a web browser, open Settings → Billing.
3. In API credits, choose Link organization and pick the Claude Console organization that should receive them, or create one. You need to be the plan’s subscriber, and an Owner, Admin or Billing member of that organization.
4. The credits appear under Promotional credits on the Console’s Billing page, with their expiry date. Create an API key and start building. No payment method is needed.

Four details that matter once you use them:³

• Choose the organization carefully. A plan links to one organization, and you can’t change it yourself; only support can.
• They expire every billing cycle. Nothing rolls over, so an unused month is money left on the table.
• Every key in the organization shares them. Give each project its own workspace with a spend limit, so one experiment can’t empty the month.
• When they run out, requests stop. Nothing is charged to your plan, unless you have bought credits or turned on auto-reload.

To stretch them, the same page suggests the obvious and the less obvious: pick the smallest model that does the task well, cache context you send repeatedly, and send work that can wait through the Message Batches API, which costs half.³

Everything else in the box

The Max plan also includes, at the time of writing:²

• Research, for questions that need many sources read and weighed.
• Memory, so Claude keeps what it learned about your work between conversations.
• Claude in Chrome, which acts in your own browser, with your sign-ins.
• Longer, multi-step tasks that keep going in the background: reports, spreadsheets, presentations.¹
• Claude Design, Slides and Docs, and Dashboards in beta.
• Claude for Microsoft 365 and Outlook.
• Skills and connectors, code execution to create and edit files, web search, projects, artifacts and voice mode.

What Max does not include is the team layer: shared projects, single sign-on, admin controls and spend controls per user belong to the Team and Enterprise plans.²

How we use ours

We looked at a month of our own use, about 90,000 requests from 22 September to 22 October 2026. Three things stood out.

• Almost everything ran on the largest model: 98.6% of what we processed went through Opus. For careful engineering that is the right call. For repetitive work, a smaller model would stretch the plan further.
• Parallel agents were about half of it. When Claude Code sends several agents to research or review at once, the work goes faster and the usage multiplies. It is the biggest lever we have, in both directions.
• The scheduled jobs were cheap: everything we run unattended came to about 4%. We still moved them to the early morning, so they never compete with the working day.

The habit that made the difference was simple: type /usage in Claude Code to see where you stand against the session and weekly limits, and look before a big job, not after.

A five-minute checklist

• Claim the API credits, and link the organization you will really build in.
• Give each project its own workspace and spend limit.
• Check /usage before long or parallel work.
• Match the model to the task: the largest for judgment, smaller for repetition.
• Turn on Research, Memory and Claude in Chrome where they save you time.
• Put this month’s credits to work before they expire.

Notes. ¹ Claude Help Center, “What is the Max plan?”. ² Claude, Max plan page, features and comparison table. ³ Claude Platform Docs, “API credits for Max and Team plans”. All read on 23 October 2026; plans and prices are subject to change. Our own figures come from our Claude Code logs for 22 September to 22 October 2026.