Analysis · August 31, 2026

Claude Max 20x vs 5x: you pay 2x, the week gives you 1.7x

Anthropic sells the $200 Max plan as “20x” Pro and the $100 plan as “5x” Pro. Read side by side, that sounds like 4x the usage for 2x the price. This weekend X did the math in public: the 4x only exists inside the 5-hour session window. On weekly limits, the numbers Anthropic itself published in 2025 put the $200 plan around 1.7x the $100 plan for Sonnet, and closer to 1.3x for Opus. The receipts below: the post that blew up, the archived support page, the lawsuit that said the same thing in June, and the September 14 change that makes the weekly cap even tighter. Then the practical part: how to keep working when Claude tells you to come back Tuesday.

The post that started it

On August 30, designer SataEric posted the math. Two million views and 7,000 likes later, “20x” was trending next to the word “misleading”.

X post by @SataEricUX with 7.3K likes: the whole 20x in Claude's pricing is so misleading. The $100 plan says 5x Pro limits, the $200 plan says 20x, but the 20x applies only to the 5-hour usage window and the weekly limit is basically 2x the $100 plan. — The post that broke the math open. View on X →

The replies are a wall of people discovering their own plan. This one sums up the mood, a quarter million views on its own:

X post by @rezoundous: wait a second, you mean to tell me Claude Max x20 is only x20 for the 5 hour session limit and not the weekly limit? — 2,600 likes for finding out. View on X →

Anthropic's own numbers, while they were public

This is not community guesswork. In 2025, Anthropic's support page for Claude Code spelled out what each plan gets per week. Here is the page as archived on September 4, 2025:

Anthropic support page archived September 4, 2025. Max 5x at $100 per month: about 225 messages every five hours, and 140-280 hours of Sonnet 4 plus 15-35 hours of Opus 4 within weekly usage limits. Max 20x at $200 per month: about 900 messages every five hours, and 240-480 hours of Sonnet 4 plus 24-40 hours of Opus 4 within weekly usage limits. — Anthropic's help center, September 2025 snapshot. See the archive →

Put the session numbers and the weekly numbers in one table and the problem is obvious:

Plan	Price	Messages / 5h	Sonnet / week	Opus / week	Weekly vs Pro
Pro	$20	~45	40–80h	—	1x
Max 5x	$100	~225	140–280h	15–35h	~3.5x
Max 20x	$200	~900	240–480h	24–40h	~6x

Source: Anthropic's published 2025 estimates. Session multiplier 5x → 20x: 4x. Weekly Sonnet multiplier: ~1.7x. Weekly Opus: ~1.3x.

Weekly Sonnet hours, by plan

Pro

$20/mo

40–80h

Max 5x

$100/mo

140–280h

Max 20x

$200/mo

240–480h

The orange bar costs twice the middle one. The “20x” names the session window, not this chart.

The detail people keep repeating: at these ranges, two Max 5x subscriptions add up to 280–560 weekly Sonnet hours. One Max 20x is 240–480. Same $200. Whether stacking accounts is allowed under Anthropic's terms is a separate question, but as a pricing signal it says everything: the top tier sells burst capacity, not weekly volume.

The numbers are gone now

Try to find those weekly hour estimates on today's help pages. We could not. The support article that carried them now describes limits without weekly numbers, and the old URL for the Max usage article redirects. The only place the figures survive is the Wayback Machine, which is exactly where X went digging:

X post by @Hesamation: Anthropic used to clarify the weekly limits for Claude Max back in July 2025. Max 5x: 140-280 hours per week. Max 20x: 240-480 hours per week, about 1.7x more weekly usage than 5x. Then they removed the numbers and never published them again. Includes a Wayback Machine screenshot with the numbers highlighted. — The archive dig, 70,000+ views. View on X →

So today a Max 20x buyer gets a multiplier that describes five hours, a weekly cap with no published size, and a /usage screen that only tells you the percentage after you have spent it.

A lawsuit said this in June

None of this is technically news. On June 15, 2026, Karl Kahn filed a proposed class action against Anthropic in the Northern District of California (Kahn v. Anthropic PBC). The complaint alleges the advertised multipliers overstate what the plans deliver once weekly caps kick in: Max 5x around 3.5x Pro, Max 20x around 6 to 8x Pro. Kahn says a single 5-hour Claude Code session burned 15% of his weekly allowance. The suit seeks to cover US consumers who bought Max since April 2025. These are allegations, not findings. Anthropic declined to comment. What changed this week is that the discourse caught up with the docket:

X post by @Hesamation: someone literally sued Anthropic over this in June. Based on Anthropic's own weekly estimates cited in the lawsuit, the 5x plan is about 3.5x Pro weekly and the 20x plan about 6x Pro weekly. You pay 2x and get about 1.7x more weekly usage. The case is ongoing, so these are allegations, not a court finding. — The lawsuit math, resurfaced. View on X →

September 14: the week gets 17% shorter

Why did a June lawsuit and a 2025 support page blow up on the last weekend of August? Because Anthropic just announced the next change to the same weekly caps:

Since May, Claude Code weekly limits ran with a temporary 50% boost. That boost ends September 13.
From September 14, the permanent level is 25% above the original baseline.
Net effect versus what you have today: minus 17%. Anthropic's own clarification: “this works out to a 17% reduction in weekly limits on Claude Code.” The first announcement led with the increase; it was deleted and reposted after the community did the subtraction.

If the weekly cap is already the wall you hit, that wall moves closer in two weeks. It applies to Pro, Max, Team and seat-based Enterprise plans.

The real cost is not the $200

Here is the part that matters once the outrage cools. When the weekly cap hits, you are not throttled, you are out. Mid-build, mid-refactor, until the reset. The instinct is to open ChatGPT or Gemini and keep going. Then you remember the new assistant knows nothing: not your projects, not your stack, not the decisions you settled three weeks ago. That accumulated context is the actual lock-in, and it is worth more than the subscription. The model you can rent anywhere. The context is yours, if you take it with you.

Keep working when Claude says come back Tuesday

This site does one thing: it turns your Claude export into a context document any other assistant understands. Three steps:

In Claude: Settings → Privacy → Export data. The ZIP arrives by email in minutes. Full export guide →
Upload the ZIP here. Every conversation is analyzed and distilled: projects, preferences, people, decisions.
Paste one document into ChatGPT, Gemini, Grok, DeepSeek or Le Chat. Your top 3 conversations are free, full migrations start at €9.99 one-time.

Move your Claude context now

For the two-assistant life

Running Claude for the hard problems and something else for the overflow is the rational setup now. AI Passport keeps one living profile of you, formatted for every assistant, re-synced as you work. Weekly caps stop being lock-outs and start being a routing decision. €9/month, founding price. Get your passport

Frequently asked questions

Is Claude Max 20x really 20x more usage?+

Per 5-hour session, yes by Anthropic's own 2025 figures: roughly 900 messages vs 45 on Pro. Per week, the same figures showed 240-480 hours of Sonnet vs 40-80 on Pro, about 6x. Compared to the Max 5x plan, the weekly gap was about 1.7x for 2x the price.

What weekly limits did Anthropic publish for Claude Max?+

In 2025, Anthropic's support pages estimated 40-80 weekly Sonnet hours on Pro, 140-280 Sonnet hours plus 15-35 Opus hours on Max 5x, and 240-480 Sonnet hours plus 24-40 Opus hours on Max 20x. Current help pages no longer list weekly hour estimates. An archived September 2025 snapshot preserves them.

What changes to Claude Code limits on September 14, 2026?+

The temporary 50% weekly limit boost that ran through the summer ends, replaced by a permanent 25% increase over the original baseline. Anthropic's own wording: this works out to a 17% reduction in weekly limits on Claude Code compared to today. It applies to Pro, Max, Team and seat-based Enterprise plans.

What is the lawsuit about Claude Max usage limits?+

Kahn v. Anthropic PBC, a proposed class action filed June 15, 2026 in the Northern District of California. The complaint alleges the advertised 5x and 20x multipliers overstate real weekly usage, putting Max 5x around 3.5x Pro and Max 20x around 6 to 8x Pro. These are allegations. There is no ruling yet, and Anthropic has declined to comment.

Can I keep my Claude context when the weekly limit locks me out?+

Yes. Export your Claude data (Settings → Privacy → Export data), upload the ZIP here, and you get one distilled context document to paste into ChatGPT, Gemini, Grok, DeepSeek or Le Chat. Your projects, preferences and decisions carry over. Your top 3 conversations are free.

Are two Max 5x plans better than one Max 20x?+

On Anthropic's 2025 numbers, two 5x plans add up to 280-560 weekly Sonnet hours against 240-480 for one 20x, at the same $200. People on X keep pointing this out. Before acting on it, check Anthropic's terms on multiple accounts. The cleaner conclusion: the 20x tier buys session burst, not weekly volume.

Sources

Anthropic help center, “Using Claude Code with your Pro or Max plan” (archived September 4, 2025)
Engadget: Anthropic hit with lawsuit over its Claude Max usage limits (June 2026)
BleepingComputer: Anthropic is cutting Claude Code's current weekly limits by 17% (August 2026)
Posts on X by @SataEricUX, @rezoundous and @Hesamation (August 30–31, 2026)

Ready to switch, or run both? The full Claude → ChatGPT guide → · Claude vs ChatGPT →