In 60 seconds: Anthropic sells Claude in four consumer tiers — Free ($0), Pro ($20/month), Max 5x ($100/month) and Max 20x ($200/month) — and none of them comes with a fixed message count. Limits are token-based: a rolling five-hour session window plus a weekly cap. Here's the full breakdown of what each plan actually gets you, and which one fits the way you work.
If you've ever Googled "how many messages does Claude Free give you," you've already hit the most confusing thing about Anthropic's pricing: the answer isn't a number. Claude doesn't sell you messages. It sells you tokens, metered across a five-hour rolling window and a weekly ceiling — and the multipliers everyone quotes (5x, 20x) only describe one of those two limits. Here's the thing: once you understand the system, picking a plan gets easy. Let's unpack it.
- Claude Free costs $0 with baseline usage per five-hour session; Pro is $20/month ($17/month billed annually); Max 5x is $100/month and Max 20x is $200/month.
- There is no fixed message count on any plan — limits depend on the model you use, conversation length and attachments, and reset on a rolling five-hour window plus a weekly cap.
- Pro adds Opus access and Claude Code; Max tiers add priority at busy times and much bigger allowances — but the same models.
- Usage is shared across the web app, desktop app and Claude Code: a heavy morning of coding eats your afternoon chat budget.
- The fine print moves often — one federal lawsuit alleges Max 20x delivers less than advertised — so check Anthropic's pricing page before you pay.
The Anthropic logo. Claude's consumer plans run from Free ($0) to Max 20x ($200/month). Logo: Anthropic, via Wikimedia Commons (public domain)
The lineup: four plans, from $0 to $200
As of October 2026, Anthropic's consumer lineup has four rungs. Free costs nothing and covers everyday questions with a baseline usage allowance. Pro is $20 a month, or about $17 a month if you pay $200 up front for the year — and it gives you at least five times Free's usage per session. Max 5x is $100 a month for five times Pro's per-session usage. Max 20x is $200 a month for twenty times Pro's. Both Max tiers are monthly-billing only; there's no annual discount, and Anthropic says it offers no standard discounted pricing on any paid plan.
For teams, there are Team seats (Standard at $20/seat/month billed annually, or $25 monthly; Premium at $100/$125 with five times the usage) and Enterprise plans priced per seat plus API-rate usage. But most readers are choosing between the four consumer tiers, so that's where we'll focus.
How the limits actually work
Anthropic's pricing page describes the deal plainly: "Every plan has usage limits that reset on a rolling five-hour session window, and paid plans add weekly limits on top." Two layers, always. The five-hour window caps how hard you can go in one sitting; the weekly cap limits your total across the week. Paid plans get both; Free gets a lighter version of the allowance.
Crucially, the 5x and 20x multipliers describe per-session usage — not your week. A Max 20x subscriber can burn through a huge amount in one five-hour sprint and still hit the weekly ceiling days later. That's exactly what a federal lawsuit alleges happened to one $200-a-month subscriber, who says he chewed through 15% of his weekly allowance in a single session. More on that below.
| Plan | Price | Per-session usage | Models | Claude Code | Best for |
|---|---|---|---|---|---|
| Free | $0 | Baseline ("covers everyday questions") | Sonnet, Haiku | No | Light everyday chat |
| Pro | $20/mo ($17/mo annual) | At least 5x Free | Opus 5.5, Sonnet 5.5, Haiku 5.5 | Yes | Daily work + coding |
| Max 5x | $100/mo | 5x Pro | Same + priority at peak times | Yes | Hitting Pro's ceiling |
| Max 20x | $200/mo | 20x Pro | Same as Max 5x | Yes | All-day heavy use |
A few things the table doesn't show. Pro's paid tier also brings higher weekly caps with a separate allowance for Opus, plus Cowork and Research features. When you do hit a limit on a paid plan, you don't always have to wait — you can buy usage credits at standard API rates as overflow, though that's real money on top of the subscription (and sales tax gets added to the plan price too). The priciest models, Fable 5 and 5.1, can use up to half of normal weekly limits on Max and premium Team seats; on Pro they run on usage credits instead. And every personal plan carries a 200,000-token context window, per Anthropic's pricing materials.
Data-center compute is the reason no AI subscription is truly unlimited — every tier rations tokens. Photo: Carl Lender / Wikimedia Commons (CC BY 2.0)
Why there's no message count
Here's why Anthropic will never tell you "50 messages a day": a message isn't a unit of cost. A short question to Haiku barely registers against your window. A long agentic coding session with Opus 5.5 — the model reading files, calling tools, re-reading context dozens of times — can burn through the same five hours fast. Same plan, same price, wildly different runway.
The other surprise: your usage pools together. Time in the Claude web app, the desktop app, Cowork and Claude Code in the terminal all draw from the same allowance. A heavy morning of agentic coding means less left for afternoon chat. During peak demand, Anthropic says a session allowance can even run out faster than five hours of clock time. None of this is a scandal — compute is finite, and a flat-price plan has to ration it somehow. But it's the number-one reason people feel blindsided.
Which plan actually fits you
Stay on Free if you use Claude the way most people do: everyday questions, drafting help, the occasional deep research dive. You get Sonnet 5 and Haiku plus core features like Artifacts and file uploads. If you rarely see a limit warning, you're not missing anything.
Pay for Pro if you hit Free's limits most weeks, or if you specifically want Opus-class models and Claude Code in your terminal. At $20 a month it's the sweet spot for daily writing, research and coding work — and the annual option drops it to about $17 a month. One honest rule from people who've done the math: if you're bumping into Pro's ceiling more than once or twice a week, that's your signal to look at Max. If it happens once a month, you don't have a Max problem — you have a one-day problem, and usage credits are cheaper.
Consider Max 5x if Pro's limits are a weekly event in your life — typically developers running long Claude Code sessions or researchers living in long-context work. You get five times Pro's per-session usage, priority when traffic is heavy, higher output limits and early feature access. Max doesn't give you smarter answers; it gives you more runway.
Max 20x is for the small group running Claude essentially all day — multiple parallel coding agents, day-long research pipelines. At $200 a month it works out to about $6.67 a day, which can undercut paying the same token volume through the raw API. Everyone else should think twice.
The fine print that keeps moving
Two honest caveats before you pull out a credit card. First, these terms change on Anthropic's schedule, not yours. Weekly limits were permanently raised 25% in September 2026 after a summer promotion ended; peak-hour session adjustments have come and gone. The numbers in this guide were checked against Anthropic's pricing page and Help Center in October 2026 — verify them again before you pay, because they will move.
Second, the "what you actually get" question is currently before a federal court. In Kahn v. Anthropic PBC, filed June 2026 in Northern California, subscribers allege the Max 5x and 20x plans deliver far less real capacity than the marketing implies once the weekly ceiling applies; an expanded class-action complaint followed in September. Anthropic declined to comment. Whatever the outcome, the lesson for buyers is the same: the multiplier describes your sprint speed, not your weekly mileage. Read both limits before you subscribe.
And if you're trying to stretch whatever tier you're on: start a new chat when you change topics (long chats burn faster), don't re-upload the same big file repeatedly, ask for short answers when you don't need long ones, and shift token-heavy background work to off-peak hours. Small habits, noticeably more headroom.
Frequently asked questions
Is Claude really free?
Yes. The Free plan costs $0 and covers everyday chat with Sonnet 5 and Haiku, with a baseline usage allowance that refills on a rolling five-hour window. You don't get Claude Code, Opus models or Cowork on Free.
How many messages do I get on Claude's free plan?
Anthropic doesn't publish a message count. Your limit is token-based and depends on which model you use, how long your conversations run and what you attach. Short chats to Haiku last far longer than long Opus sessions under the same allowance.
Is Claude Pro worth $20 a month?
For daily users, usually yes: at least five times Free's per-session usage, Opus access, Claude Code, higher weekly caps and usage credits instead of dead ends. Light users who rarely hit Free's limits should stay put.
What's the difference between Claude Max 5x and Max 20x?
Only volume and price: Max 5x costs $100/month for five times Pro's per-session usage; Max 20x costs $200/month for twenty times Pro's. Features are identical — same models, priority access and early features on both.
Do the limits reset?
The five-hour session window refills on a rolling basis, and the weekly cap resets once a week. Pro and Max subscribers can also buy usage credits at API rates to keep working past a cap.
Related reading: our OpenAI API rate limits explainer (RPM, TPM and 429 errors), ChatGPT token limits in 2026, and Anthropic's $2 trillion IPO papers.
Sources: claude.com/pricing and the Anthropic Help Center, via independent guides that checked them in October 2026 — threatfrontier.com (checked Oct. 11), aionx.co (checked Oct. 15), dev.to, klymentiev.com, magai.co, izzedo.chat; startupfortune.com and The Verge on Kahn v. Anthropic PBC. Prices and limits change — verify on Anthropic's official pages before subscribing.