How much does Claude Pro cost?

About $20 a month, or around $17 a month if you pay annually. Claude Max tiers start at $100 a month for much higher capacity (Anthropic pricing page, as-of 2026-08-29).

What happens when you hit the Claude Pro usage limit?

Your asking stops until the limit resets. Heavy users reach that point regularly, which is why the tiers above Pro start at $100 a month.

Is Claude Max worth the money?

Only if you use the frontier model most days for hard work: complicated code, long documents, drafts that must be right first time. For everyday questions and drafting, most people pay for capacity they rarely call on.

Is Claude better than ChatGPT for writing?

Both frontier models write well. Claude is the one people reach for on long-form work: a report, a chapter, a document that has to hold its structure across thousands of words.

Is there a cheaper alternative to Claude Pro?

Yes, if your questions are ordinary ones. Plainly costs $5 a month and automatically uses the best available model for your question, which gives far more asking than a frontier plan's limits allow, at a fraction of the price.

Claude Pro usage limits: what runs out

Claude Pro costs about $20 a month and its limits arrive in practice rather than on paper: heavy use runs out, and Max starts at $100. Who should not pay.

Claude Pro usage limits

Claude Pro's usage limits are real. About $20 a month buys frontier access with a ceiling, heavy use reaches that ceiling before it resets, and the tier above starts at $100 a month.

What the Claude tiers cost

Anthropic's pricing page (as-of 2026-08-29) lists three levels.

Plan	Price	What you get
Free	$0	The smaller model, capped
Pro	~$20 a month, or ~$17 a month paid annually	Frontier access, with usage limits
Max	From $100 a month	Much higher capacity

The entry price reads like ChatGPT Plus, which sits at $20 a month. The difference is what happens after you pay it. Pro isn't the tier a heavy user stays on. The limits arrive, and the next rung up costs five times as much. A cheap entry price with a working price several times higher behind it's the shape of every frontier vendor's ladder, and it's the ladder set out in free vs $5 vs $20 AI apps.

What running out looks like

Heavy users don't stay under the ceiling. Ask questions all day and you'll reach it.

Anthropic's usage-limit documentation (as-of 2026-08-29) describes two kinds of limit. One is a rolling session window: roughly a five-hour stretch with a small amount of usage inside it. The other is a weekly cap: a larger amount, but a ceiling on the week. Hitting either one stops the asking. When Claude reaches a limit mid-task it simply stops working until the limit resets, and it doesn't let you finish what you were doing.

I ran out on Pro more than once. The ceiling arrived during long, ordinary sessions rather than one heroic task. The tier is priced around ordinary use and sold as a capable one. I believe that is deliberate, and it's why the tiers above exist at all.

The $100 a month

For a period I paid as much as $100 a month for AI capacity. It worked. Nothing I asked for was refused, nothing was cut off halfway, and the ceiling stayed out of the way. It was also overkill. I was paying for the heaviest fortnight of the year and using a small share of it in an average one.

The realisation

The extra $95 bought nothing I could point to for the questions I actually had. Open-source models now sit slightly off the frontier. They're very good at everyday work: questions, planning, drafting, explaining something back to you. An app that uses the best available model for each question, chosen for you, with no ceiling on how often you ask, costs $5 a month. That's $95 less for more use, and the arithmetic behind it's the subject of how a $5 AI app makes money.

When the frontier tier is still the right buy

Two readers, two answers.

Stay on Pro, or step up to Max, if the work is frontier work most days: complicated code across a codebase, specialist research through long documents, drafts that have to be right first time, long-form writing that gets published. That work earns the price, and the case for paying it's set out in is ChatGPT Plus worth it.

Move to a $5 tier if your questions are everyday ones. Most are. An email, a plan, a term in a contract, a second opinion on a decision: none of those need the frontier, and at $5 the asking doesn't stop. The long case for that side of the line is in is a $5 AI subscription good enough and what a cheap ChatGPT alternative is.

In my experience, Claude is the strongest writing tool I've used. Long documents, a careful tone, a draft that has to hold together across thousands of words: that is where it wins, and where the gap to the cheaper models is still widest. I believe writing quality is the last advantage the frontier keeps, and it's a real one.

Where this comparison falls short

Two limits on the argument above.

The first: a $5 app isn't a frontier subscription, and it doesn't pretend to be. You won't get the absolute frontier models, and on the hardest problems the difference is real. If your first draft has to be right, or the model has to work through something hard, pay for the frontier. The ceiling is what the top tier is for, and no cheap app gives it to you.

The second: every figure here is dated 2026-08-29 and AI prices move. Claude's tiers have changed before and will change again. Check Anthropic's pricing page before you decide, and treat the shape of the argument as the part that lasts. A cheap entry price sits on top of a steep working price, and underneath both there is a capable alternative at $5.

What to do next

If you're on Pro and running out every month, spend one month doing your ordinary asking on a $5 app. Then ask whether you missed the ceiling above it. If you did, the premium tier was earning its price and you should go back. If you didn't, the limits were the only thing you were paying for.