Pro is $20. Max 5x is $100. The only number that matters is how many times a week you hit the five-hour wall.
Both plans include Claude Code and the same apps. Max buys more of the same allowance, and Anthropic has never published how much of it either plan holds.
The two plans side by side
From claude.com/pricing and the Anthropic Help Center, read 09/28/2026. Prices exclude tax.
| Pro | Max 5x | Max 20x | |
|---|---|---|---|
| Monthly price | $20 | $100 | $200 |
| Annual option | $17 a month ($200 up front) | None, monthly only | None, monthly only |
| Usage per five-hour session | At least 5x Free | 5x Pro | 20x Pro |
| Claude Code | Included | Included | Included |
| Price per unit of Pro usage (monthly billing) | $20 | $20 | $10 |
The last row is the useful one. Max 5x costs five times Pro for five times the usage, so the rate is the same. Max 20x costs ten times Pro for twenty times the usage, so it halves the rate. You only reach the discount if you use it.
What neither plan tells you
No Claude plan states a number of tokens or messages. Anthropic describes each allowance as a multiple of another plan, and its usage-limits article says the real amount depends on conversation length, features, the model and the effort level.
Any article that says “Max equals $X of API usage” is estimating, because the allowance it is comparing against is not public. I will not give you that number. What I can give you is the API side, measured, and a rule for the plan side that you can check against your own week.
Both plans also share one pool across every Claude app. Anthropic’s help article states that claude.ai, Claude Code and Claude Desktop all count toward the same limit. If you code in the morning and chat in the afternoon, both come out of the same allowance. The details are in Claude usage limits, explained.
How to choose between Pro and Max
Use your own limit hits as the measurement.
- You rarely hit the five-hour limit: stay on Pro. Max buys headroom you will not use.
- You hit it a few times a month: stay on Pro. Waiting for a reset or buying usage credits now and then costs less than $80 a month.
- You hit it several times a week, and the wait stops your work: move to Max 5x.
- You still hit it on Max 5x: move to Max 20x. It is the only monthly plan where each unit of usage costs less.
In Claude Code, /status shows where you are against the limit. Check it at the end of a normal day for a week before you upgrade. If most days end with room to spare, the problem is a few heavy sessions, and I think the ten habits in Claude Code usage limits fix most of it for free.
Max vs the API
The API bills every token at a listed price, so its side of the comparison can be measured exactly.
I measured one full day of my own Claude Code work: 2,979 turns of agentic coding with a large context carried on each turn. In Opus 5.5 vs Fable 5.1 I replayed that day at the current list prices of several models. Two of them:
| Model | API cost of that one day |
|---|---|
| Opus 5.5 | $220 |
| Opus 5 | $359 |
One heavy day on the API costs more than a month of Max 20x. I cannot tell you whether Max 20x would have covered that whole day, because the allowance is not published. But the size of the gap explains why people who run Claude Code all day choose a plan.
The comparison flips for light use. Suppose you send 40 requests each working day through the API, each about 2,000 tokens in and 500 out, on Sonnet 5.5 at $2 per million input tokens and $10 per million output. Over 21 working days that is 840 requests: $3.36 of input and $4.20 of output, or $7.56 a month. That is well under Pro. The only way to know which side you are on is to price your own work.
How to price your own day
- Take a typical prompt from your work, including the context you paste or attach.
- Paste it into the Claude token counter and pick the model you use.
- Multiply the per-request cost by how many requests you send in a normal day, then by about 21 working days.
If the result is well under $20, the API is cheaper than Pro. If it is well over $200, I would bet on a plan, even though the exact allowance is unknown. The middle is where the honest answer is “try Pro for a month and watch /status“.
The person who counts their own limit hits for a week is way ahead of the one who guesses.
The full price list for every model is in Claude pricing.
When prices change
API prices have moved in recent weeks, and Anthropic edits its limit articles. A change to either shifts where your break-even sits.
I am building Claude Cost Watch, a short email that tells you when a Claude price or limit changes. It is a waitlist today, and nothing is charged.
Sources
All read on 09/28/2026.
- Plan prices and usage multiples: claude.com/pricing
- Weekly limit and five-hour reset: Anthropic Help Center, “What is the Max plan?”
- Shared limit across apps: Anthropic Help Center, “How do usage and length limits work?”
- API prices: platform.claude.com/docs/en/about-claude/pricing
- Measured day: Opus 5.5 vs Fable 5.1
No link on this page is an affiliate link. Anthropic has no affiliate program I could verify.