Claude Max x5 and Max x20 Pricing and Subscription Plans Features for August 2026

How much is Claude Max in total, and how does it sit against the tiers people weigh it against? Every Anthropic plan is below. The Claude Max price has not moved since the tiers launched: $100 for 5x and $200 for 20x.

Anthropic sells Max as an individual plan only. It is not available for teams, it cannot be billed annually, and unlike Pro there is no discounted yearly option. Prices below exclude tax. Buying through the iOS or Android app can cost more than buying on the web, so check both before you subscribe.

Upgrades are prorated and apply immediately. Downgrades take effect at the end of the current cycle and your projects and chats survive the change. Switching from an annual Pro plan to Max can credit the unused balance to your account, provided the billing addresses match.

Claude Max limits are the part almost every guide gets wrong, because almost every guide describes a single usage allowance. There are two, they run simultaneously, and understanding the difference is what separates people who are happy on Max 5x from people who cancel Max 20x in frustration.

The rolling five-hour session window

Your session clock starts on your first message, not at midnight. It advances on a rolling basis, so exhausting it means waiting for the window to move rather than waiting for a daily reset. This is the quota the 5x and 20x multipliers apply to.

The two weekly caps

Sitting above the session window are two separate weekly limits: one across all models combined, and a second capping Sonnet specifically. Both reset at a fixed time tied to your account, visible under Settings then Usage. There is no rollover. Anthropic introduced these in August 2025 and states they affect fewer than 5% of subscribers.

Everything draws from one pool

Web chat, the desktop app, mobile and Claude Code all consume the same allowance. Claude Code is not a separate quota. It also loads roughly 20,000 tokens of repository context at the start of a session, which means a terminal workflow can eat a meaningful slice of a “generous” limit before you have typed a real prompt.

What burns quota fastest, in order: long conversation context, Opus and Fable over Sonnet and Haiku, resuming an old chat rather than starting fresh, attachments, Research and tool calls, extended thinking, and multi-file Claude Code sessions.

Approximate capacity per five-hour window

Anthropic does not publish message counts. The figures below come from independent testing and user reports, and the variance between users is genuinely large because it depends on which model you run and how long your context gets.

Read the Opus column again. Max 20x costs exactly double Max 5x, and its weekly Opus ceiling is roughly 24 to 40 hours against 15 to 35. That is not a 4x improvement. It is not even reliably a 2x improvement.

This is the single most common cause of cancelled Max subscriptions, and it is not a secret so much as an unstated detail. The 5x and 20x figures describe how much you can consume inside one rolling five-hour window. They say nothing about the weekly caps that sit above it.

Users who have run both tiers put the real weekly difference between Max 5x and Max 20x at roughly 1.5x to 2x. One subscriber reported that Anthropic’s own support confirmed the weekly increase is about 1.7x, not the 4x the price difference suggests. Another cancelled 20x after finding the weekly ceiling arrived sooner than expected and described the advertised multiplier as misleading.

The practical consequence is a clean decision rule. Work out which wall you actually hit.

Check Settings then Usage for a fortnight before upgrading. The graph tells you which wall you are hitting, and that answer decides the plan more reliably than any review.

Claude Pro vs Max is the most searched question in this whole space, and the answer is unusually clean. Framed either way round, Claude Max vs Pro comes down to a single variable. Pro and Max run the same models, expose the same features and share the same 200K context window. Nothing about the quality of the answer changes when you pay $100 instead of $20.

What users actually report is that the jump feels larger than 5x, because removing interruption changes how you work rather than just how much you can send. Several describe using Opus for 90% of their work on Max 5x when that was impractical on Pro. Others working eight-hour days on Max 5x say they never exceed 25% of the weekly cap.

The dissenting reports matter too. Some Max 5x subscribers describe hitting the cap after five to ten lightweight prompts with all extensions disabled, and note that resuming a long chat can consume 10 to 15% of a five-hour window before the first new answer. One user burned 76% of a session limit and 7% of a weekly limit on a single prompt inside a very long conversation. The variance is real, and context length explains most of it.

Claude Max 20x costs exactly twice Max 5x. On the session window the multiplier is genuinely 4x relative to 5x. On the weekly cap it is closer to 1.7x. Which of those two numbers matters to you depends entirely on whether your work is deep or wide.

Which produces an arbitrage worth knowing: one Max 20x account costs the same per month as two Max 5x accounts. If you run parallel isolated projects, two 5x accounts give you two independent session windows and two independent weekly caps for the same $200. If you need one enormous unbroken session, that split helps you not at all and 20x is the answer.

The community pattern for anyone unsure: start on Max 5x for one month. Upgrades are prorated and instant, so moving to 20x mid-cycle costs you nothing in wasted spend if 5x turns out to be insufficient.

“Is it worth it” is not a matter of opinion once you know roughly how many tokens you consume in a month. Compare what that volume would cost at standard API rates against the flat subscription price, and the right tier picks itself.

Heavy Claude Code users routinely report Max returning five to seven times its sticker price in equivalent API value. That is the strongest argument for the plan, and it only applies if your volume is genuinely sustained rather than bursty. People whose usage spikes for a week and then goes quiet tend to cancel, because a flat $100 or $200 has to be earned every month.

There is no Claude Max discount currently on offer. Anthropic states plainly that no standing discounts exist for Max, and there is no trial on either tier. If you want to test Max, you pay for a month. The free tier remains available indefinitely, and Pro is the cheaper way to find out whether limits are actually your problem.

The one genuine exception on record: in February 2026 Anthropic ran a Claude for Open Source programme offering six months of free Max 20x access to qualifying open-source maintainers, with no automatic rollover to a paid plan afterwards. That is the source of most searches for Claude Max 6 months free, and for Claude Max free for 6 months. It was a targeted programme with eligibility requirements, not a general promotion, and it is not a standing offer.

Occasional promotions do appear through official channels. Anything else claiming to hand out free Max accounts should be treated with the suspicion it deserves. If cost is the constraint, the honest options are Pro at $17 a month billed annually, or pay-as-you-go usage credits at API rates on top of a cheaper plan rather than a tier upgrade.

The reported savings below are large enough that they can move you a whole tier down. Try them for a fortnight before paying five times more.

One more structural note: Opus and Fable burn quota substantially faster than Sonnet and Haiku. A common pattern among heavy users is to run Opus only for planning and orchestration, then hand execution to cheaper models. Several people describe staying comfortably inside Max 5x on full-time work purely by doing this.

Team Premium runs $125 to $150 per seat per month and requires between 2 and 150 seats plus organisational billing. Max 5x and Max 20x are $100 and $200 flat with no seat minimum. For a single power user in isolation, Max wins on price.

Team becomes the right answer when you need what Max structurally cannot offer: single sign-on, central billing, admin controls, and no-training-by-default guarantees written into the plan rather than toggled in a setting. Those are organizational requirements, not usage requirements, and no amount of individual capacity substitutes for them.

Last verified 25 July 2026 against Anthropic’s official pricing and Help Center pages. Next refresh due 25 August 2026.
Anthropic adjusts usage limits and plan structure over time and does not publish exact message counts. Every capacity figure on this page is community-reported and approximate.
Full plan-by-plan detail including API rates on the Claude pricing page.