Claude Max Plan: $100 Price, Limits, API Credits

The Claude Max plan costs $100 a month, billed monthly only. It gives 5x the Pro usage every five hours and every week, priority access, Fable for up to half your weekly limit, and $100 a month in API credits for the Claude Platform. If you want more than that, Ultra is $200. Prices checked 2026-10-31.

Prices checked 2026-10-31

Source for every number here: claude.com/pricing and Anthropic's Max, Ultra and API-credit help articles. Prices exclude tax and may change at Anthropic's discretion.

Max is the middle of three individual paid tiers. See Claude pricing for all of them side by side, and Claude Pro pricing for the step below.

What Max is for

Anthropic describes Max for people who "collaborate with Claude frequently and need more usage to work on a variety of tasks." Its listed benefits are more usage, fewer interruptions, priority access, Claude Code, longer multi-step background tasks, and monthly API credits.

In practice there are three reasons people buy it.

1. You code with Claude Code most of the day. Claude Code and chat share one usage pool, so a heavy coding day on Pro runs into the weekly limit.
2. You want Fable. On Pro it needs usage credits. On Max it is part of the plan, up to half your weekly limit.
3. You build with the API and want a monthly credit to do it. That is $100 a month toward the Claude Platform.

If none of those fits, Pro is cheaper and probably enough. Check how often you actually hit limits before moving up.

How the limits work

Max has three limits.

• A five-hour session limit, 5x Pro. The clock starts at your first message.
• A weekly limit across all models, 5x Pro. It resets at a fixed time assigned to your account.
• A separate weekly limit for Fable.

All Claude surfaces share them: claude.ai, desktop, mobile and Claude Code. Anthropic gives no message count, because usage depends on the model, chat length, file size, tool use and effort level. It also says it may adjust limits to manage capacity, so 5x is a ratio to Pro, not an absolute promise.

A Max subscriber who hits the five-hour limit mid-response in Claude Code can get a brief wrap-up allowance to finish the step in progress. On Max it can apply at each five-hour limit, within your weekly limit. It counts toward your weekly total and needs Claude Code 2.1.277 or later. Anthropic says it is rolling out gradually.

Anthropic's Max help article also notes that if you have a limit reset, you can use it to refill your five-hour session limit or weekly limits to full.

When you run out: usage credits

With usage credits enabled, the first time you exhaust any included limit (five-hour, weekly or Fable weekly), requests move to pay-as-you-go at standard API rates and keep going. The banner may say "approaching your limit" briefly, then "Now using credits." You prepay a balance, set a monthly cap or none, and can turn on auto-reload.

You can instead switch to a Console account for a heavy coding sprint, upgrade to Ultra, or wait.

Fable on Max

Fable is Anthropic's top model, $10 in and $50 out per million tokens on the API. On Max you can spend up to 50 percent of your weekly usage limit on Fable models without extra cost. Anthropic is explicit that this is not a bonus: "you can never use more than your weekly limit." Fable uses the weekly limit faster than other models. Past the Fable cap you can keep going on usage credits or switch models.

The earlier promotion that put Fable 5 into Pro plans ended on 19 July 2026. Fable 5 and 5.1 now work the same way on every plan. Neither is the default in Claude Code; you pick one with /model. Fable 5.1 needs Claude Code 2.1.255 or later.

For the difference between Fable and the cheaper tiers, see Claude models and Opus vs Sonnet.

The monthly API credit

Max includes $100 a month in credits for the Claude Platform. Ultra includes $200. Pro does not. The details come from Anthropic's help article.

What you can spend them on: the Claude API (Messages and Batch), the Console Playground, Claude Managed Agents and the Claude Agent SDK.

What you cannot spend them on: interactive Claude Code in the terminal, IDE, desktop or web; usage credits in Claude, Claude Code or Cowork; and Claude on Bedrock, Google Cloud or Foundry.

How to claim:

1. Be on Max for seven days.
2. In claude.ai, open Settings, Billing and find the API credits section.
3. Choose "Link organization" and pick or create a Claude Console organization. You need the Owner, Admin or Billing role there.
4. Accept the Supplemental Credit Terms.

You need no card on file at the Console to claim or use the credit.

Rules that bite:

• You can link only one Console organization, and you cannot change it yourself afterward. Contact support to change it.
• Each Console organization can take credits from only one plan.
• Credits refresh each billing cycle and do not roll over. Unused credits expire at the end of the cycle.
• Everyone with an API key in the linked organization draws from the same balance.
• When credits run out and there are no purchased credits or auto-reload, API requests stop until next month's credits arrive. Nothing is charged to your Claude plan.
• If you cancel or downgrade, new credits stop but existing ones stay usable until they expire.

One catch concerns scripted Claude Code. The help article says the credits cover claude -p and the Agent SDK when you run them yourself with an API key from your linked organization. When you are signed in with your plan instead, they draw from plan limits, and runs started by the GitHub Action, an IDE extension or the desktop app count as Claude Code usage that credits do not cover.

Billing mechanics

• Monthly only. There is no annual Max.
• Web price. $100 is for web subscriptions. App store pricing varies.
• Pro to Max. Limits apply immediately. If you were on annual Pro and the unused balance exceeds the Max price, you get an account credit for the rest, applied to future subscription costs. The billing address must match the earlier purchase. If it does not, Anthropic says to cancel Pro, wait for its end date, then sign up for Max.
• Google Play. If you subscribed that way you pay full price on upgrade; the unused value of the old plan converts into extra days, so renewal lands a few days later.
• Max to Ultra. Charged prorated for the rest of the cycle. If you had claimed API credits, you get prorated credits right away and the full $200 each cycle after.
• Discounts. No standard discounts on Pro, Max or Ultra. Support cannot issue one-off coupons. Occasional promotions are announced publicly.
• Refunds. Payments are generally non-refundable except where the Consumer Terms or local law say otherwise.

Three scenarios, worked from the published prices

These are arithmetic on listed prices, not measured usage.

A solo developer who bursts. Heavy two weeks a quarter, quiet otherwise. Annual Pro is $200, and a $50 usage-credit spend in each of four heavy fortnights adds up to $200 more, so $400 a year. Max would cost $1,200 a year. Pro with credits wins unless the bursts become routine.

A full-time developer on Claude Code. Hits the weekly limit most weeks on Pro. Max is $100 a month, flat. The alternative is usage credits at API rates: Anthropic's enterprise average of $150-250 a month for API-billed Claude Code sits above that. Max looks cheaper, assuming the 5x allowance holds your load.

A solo founder building an AI product. Uses Claude Code daily and spends about $100 a month on API calls in development. Max includes $100 of Platform credit, which can cover that spend if the credit is linked to the right Console organization, so the plan is partly offset. Ultra would add another $100 of credit for $100 more per month, which is break-even only if you would spend it anyway.

Max or Ultra?

Ultra replaced the old Max 20x plan; existing Max 20x subscribers moved to Ultra at no price change. Ultra accounts are verified automatically from account and payment history. An unverified Ultra account has a temporary five-hour limit until verification. Check Settings, Usage: if there is no "Current session" bar, your account is verified.

Max or API billing?

Compare on one number: your usual monthly API-rate spend for the same work. Anthropic's costs doc puts enterprise averages at $150-250 per developer per month on API billing. A $100 Max plan beats that, if 5x Pro covers you. If your work is automated or bursty, API billing may suit better. See Claude Code pricing for the working and Claude API pricing for rates.

Our take

Max is a plan for people who already know they hit Pro's ceiling. Because $100 buys 5x of a $20 plan, the price of usage does not drop; the value is fewer interruptions plus two perks Pro lacks: Fable inside the plan and a real API credit.

The API credit is the part to plan around. If you build with the API, $100 a month of Platform credit takes much of the sting out of the plan price, but only for work on the Platform and only if you link the right Console organization the first time. If you never touch the API, treat that credit as zero.

Before buying Max, run your current Pro usage for a month with a low usage-credit cap and see what you spend. If credits would have exceeded $80 across the month, Max pays for itself in avoided interruptions. If they stayed under $30, stay on Pro. We have not tested usage on either plan, so this is arithmetic from published prices, not a benchmark.

Compare with other tools at Cursor pricing, Codex pricing and the AI coding agent pricing comparison.

Checklist before you upgrade

• Look at Settings, Usage to see how close to the weekly limit you really run.
• Decide whether Fable matters to you. If not, a big part of Max's extras are moot.
• If you build with the API, pick the Console organization you want credits in before you link.
• Set a monthly usage-credit cap so a long sprint cannot surprise you.
• Remember there is no annual discount on Max.
• If you pay through an app store, note that usage credits can only be turned on at claude.ai.

Practical tips for Max subscribers

Check Settings, Usage at the start of each week to see the weekly bar, and note your reset time, which stays fixed. Plan the heaviest work for the day after a reset rather than the day before one. Keep Fable for the tasks that need it, because it draws from the weekly pool faster than other models and has its own separate weekly limit.

In Claude Code, run /usage to see which skills, subagents, plugins and MCP servers are consuming the most, and switch off the ones you do not use. Anthropic's costs documentation suggests clearing context between unrelated tasks, using Sonnet for most coding, keeping subagents on Haiku where the work is simple, and keeping agent teams small.