Pricing & Plans
Claude pricing
Every subscription tier and every per-token rate in one place, kept current on a visible cadence. If you only need the number, it is in the first table.
Claude is sold two ways that people constantly conflate. Subscriptions (Pro, Max, Team, Enterprise) buy you a usage allowance inside the apps — you never see a token bill, you see a limit. The API bills per million tokens with no allowance at all. Claude Code and Cowork sit across both, depending on how you authenticate.
That distinction decides which number matters to you. If you are on a subscription, the per-token rates are useful only as a proxy for how fast you will hit a usage limit. If you are on the API, the subscription table is noise.
| Tier | Indicative price | Billed as | Who it suits |
|---|---|---|---|
| Free | $0 | — | Trying it out; nothing to optimize |
| Pro | ~$20 / month | Subscription | One person, daily but moderate use |
| Max (lower) | ~$100 / month | Subscription | Heavy individual use, Claude Code |
| Max (higher) | ~$200 / month | Subscription | All-day agentic work |
| Team | ~$30 / seat / month | Subscription | Small teams needing shared admin |
| Enterprise | Custom | Contract | SSO, procurement, compliance |
| API | Per million tokens | Usage | Products and automation |
Rates by model — Opus-class, Sonnet-class and Haiku-class — differ by more than an order of magnitude, which is why model choice often beats prompt optimization as a cost lever. Those are broken out on the API pricing page.
In this hub
- API pricingPer-million input and output rates by model, with worked examples.
- Max planWhat the two Max tiers cost and when either one pays for itself.
- Team pricingPer-seat cost, the minimum seat count, and what admin actually gets you.
- Pro vs MaxThe upgrade decision, answered with usage patterns rather than feature lists.