The three plans, side by side
| Pro | Max 5x | Max 20x | |
| Price | $20/mo ($17 annual) | $100/mo | $200/mo |
| Claude Code prompts per 5 h | ~10–40 | ~50–200 | ~200–800 |
| Sonnet hours per week | ~40–80 | ~140–280 | ~240–480 |
| Opus hours per week | none in Claude Code | ~15–35 | ~24–40 |
| Over the limit | Wait for the reset, or turn on usage credits at API rates (up to $2,000/day) | ||
Ranges are Anthropic's own; the low end assumes big contexts and long sessions, the high end short, focused prompts. Weekly limits are shared with Claude chat and Cowork on the same account.
How to read your result
The calculator turns your hours into two numbers. The first is weekly hours, weighted so an Opus hour counts about 2.5× a Sonnet hour — that's roughly how much faster the shared weekly budget drains on Opus, and it's the number that decides whether you'll see "You've hit your weekly limit" on Thursday. The second is prompts per 5-hour window, from your session intensity, which decides whether you'll hit the rolling 5-hour cap on a busy afternoon. A plan is a fit when both numbers sit inside its range with room to spare; "tight" means you'll live near the ceiling and should expect the occasional wait or usage-credit charge.
The API line is the same workload priced per token. Anthropic's benchmark across enterprise deployments is about $13 per developer per active day and $150–250 per month, with 90% of users under $30 a day; the calculator scales that by your intensity and Opus mix. For most full-time users the API costs more than Max and adds rate-limit and billing overhead, which is why the subscription exists. The API wins when you use Claude Code lightly, need it in CI or automation, or run workloads a subscription can't legally share.
Where people go wrong
- Buying Pro for all-day work. Pro's 40–80 Sonnet hours a week sounds like a lot until a long session with a big context burns through it in three days. If Claude Code is your main editor, start at Max 5x.
- Buying Max 20x for Opus. Opus hours per week are only modestly higher on 20x (24–40) than 5x (15–35). 20x is about volume — parallel sessions, agents, 8-hour days — not about Opus access.
- Forgetting chat usage. Every Claude.ai conversation, Cowork task and Claude Design session on the same account draws from the same 5-hour and weekly pools.
- Ignoring context size. Claude Code resends your whole conversation with every request. A one-line question in a session that's been open all day costs as much as the whole history.
/clearbetween tasks is the cheapest upgrade there is. - Assuming the API is always cheaper. It is for light use. At 2+ hours a day it usually isn't, and you lose the 1-hour prompt-cache lifetime subscribers get.
Annual vs monthly, and Team seats
Pro is $17/month billed annually ($200 up front) vs $20 monthly; Max is monthly only. If you're buying for a company, Team standard seats are $25/month ($20 annual) with more usage than Pro, and Team premium seats are $125/month ($100 annual) with about 5× standard usage — roughly Max 5x with admin controls, SSO and pooled billing. Enterprise is annual, per-seat plus API-rate usage.
Will I hit the weekly limit? →
Frequently asked questions
Is Pro enough for Claude Code?
For 1–2 hours a day of normal work on Sonnet, usually. All-day use, agents, or Opus need Max.
Max 5x vs Max 20x?
5x: ~50–200 prompts/5 h, 140–280 Sonnet + 15–35 Opus hrs/wk. 20x: ~200–800 prompts, 240–480 Sonnet + 24–40 Opus hrs. 20x is for volume and parallel sessions.
Is the API cheaper?
Only for light use. Anthropic's benchmark is ~$13/developer/active day (~$150–250/mo), more than Max 5x.
What happens at the limit?
You wait for the 5-hour or weekly reset, or enable usage credits at API rates.
Does chat usage count against Claude Code?
Yes — one shared pool per account across chat, Cowork and Claude Code.
Can I switch plans mid-month?
Yes; upgrades are prorated and take effect immediately.