Claude Pro vs Max 5x vs Max 20x: which Claude plan fits your usage
Claude Pro costs $20, Max 5x $100 and Max 20x $200. What the multipliers mean, why Max 20x is the cheapest per unit of usage, and when the API costs less.
Claude Pro costs $20 a month, Max 5x costs $100 and Max 20x costs $200. Anthropic defines the Max plans by multiplying Pro’s allowance: Max 5x gives five times Pro’s usage per session, Max 20x gives twenty times. That makes Max 5x the same price per unit of usage as Pro, and Max 20x half the price per unit. If you regularly run out of Pro, the question is not whether Max is “worth it” but how many times Pro’s allowance you actually use.
The plans
| Plan | Price a month | Usage, as Anthropic describes it | Usage per dollar, Pro = 1 |
|---|---|---|---|
| Free | $0 | “Free covers everyday questions” | – |
| Pro | $20 ($17 billed yearly) | At least 5x the Free plan’s usage per 5-hour session | 1 |
| Max 5x | $100 | 5x Pro’s per-session allowance | 1 |
| Max 20x | $200 | 20x Pro’s per-session allowance | 2 |
The last column is simple arithmetic: Max 5x gives five times the usage for five times the price, while Max 20x gives twenty times the usage for ten times the price. Annual billing lowers Pro to $17 a month; Anthropic’s Max support page lists the Max prices for web subscriptions and notes that mobile prices may differ.
How the limits work
Claude’s usage is counted per 5-hour session. On the Max plans Anthropic also applies a weekly limit across all models. Anthropic does not publish a fixed number of messages for any plan, because a message’s cost depends on how long it is, how much of the conversation it carries, which model answers and whether tools such as Claude Code run long tasks.
That last point matters most. A short question uses a sliver of a session; an agent that reads a codebase and edits files for an hour can use a large share. Long agent sessions, not chat, are what usually push a plan to its limit.
Plan or API?
The alternative to a plan is paying per token on the API. Anthropic’s API prices per million tokens are $2 input and $10 output for Claude Sonnet 5.5, and $4 and $20 for Claude Opus 5.5, with cache hits at 5% of the input price.
A worked example shows the scale. Take a coding agent making 100 requests a day, each sending 40,000 tokens of context, 70% of it served from cache, and writing 2,000 tokens back:
- On Sonnet 5.5: 12,000 × $2 + 28,000 × $0.10 + 2,000 × $10, per million, is about $0.047 a request, or about $140 a month.
- On Opus 5.5: the same workload is about $0.094 a request, or about $281 a month.
These figures leave out cache writes, which Anthropic bills separately ($2.50 per million tokens for a 5-minute cache on Sonnet 5.5), so the real API bill would be somewhat higher. The workload is an illustration, not a measurement; your own sessions may be lighter or much heavier. At this level of use, a Max plan’s fixed price is attractive. For lighter use the API can easily cost less than Pro’s $20.
How to choose
- Pro: you chat daily, write and analyze documents, and only occasionally hit the session limit.
- Max 5x: you hit Pro’s limit most days and need to keep working, but your heavy use comes in bursts.
- Max 20x: you run coding agents for hours most days. It is the best value per unit of usage, provided you use enough of it.
- API: you use Claude inside your own app or scripts, or your usage is light and irregular. You pay only for what you use and get the full 1M-token context at standard prices on Sonnet and Opus.
Put your own numbers into the subscription or API tool, or price an agent workload in the API cost calculator.
Sources
- Anthropic, Claude pricing.
- Anthropic, What is the Max plan?
- Anthropic, API pricing.
All pages read on October 26, 2026.
Free toolIs a monthly AI plan cheaper than paying per use?Compare for my usage
