Claude’s pricing looks simple from a distance — free, then $20 — and gets genuinely complicated up close, because Anthropic sells to three different customers at once: casual users, power users who burn through limits, and organisations buying seats. This guide walks through each plan as listed on Anthropic’s pricing page in October 2026. Prices change, so treat the figures as a map and confirm the live numbers before you buy.
The first thing to understand is that subscriptions and the API are separate products. A Claude subscription buys you the assistant in its apps. The API, billed per token, is for developers building on the models, and no subscription includes a general API allowance beyond specific features like Claude Code.
Free and Pro: the individual plans
The Free plan costs nothing and covers everyday chat in the browser and apps, with web search, file uploads and the standard model, subject to usage limits that reset through the day. It is a real product, not a demo, but heavy users will meet the ceiling quickly.
Pro is $20 a month, or $17 a month billed annually. It raises usage limits substantially, unlocks the stronger models and Projects, and — the detail that matters most in 2026 — includes Claude Code, Anthropic’s coding agent. For a single person using Claude daily for writing, research or code, Pro is the plan the lineup is designed around.
Max: for people who hit the ceiling
Max exists because Pro’s limits are a real constraint for agentic work. Max 5x costs $100 a month and multiplies Pro’s usage by five; Max 20x costs $200 a month and multiplies it by twenty, with higher output limits and priority during busy periods.
These prices look steep for a chatbot and make sense for a tool: Max subscribers are typically developers running Claude Code for much of the working day. When a coding agent’s output is billable work, $200 a month is compared against an engineer’s hour, not against Netflix.
Team and Enterprise: buying seats
Team plans are sold per seat for groups of two or more. A Standard seat is $25 per person monthly, or $20 billed annually, and adds shared workspaces, central billing, single sign-on and admin controls. A Premium seat — $125 monthly or $100 annually — pairs the Team workspace with Max-class usage, aimed at engineers on the plan who live in Claude Code. Standard and Premium seats can be mixed on one bill.
Enterprise is the negotiated tier: listed from $20 per seat as a platform fee, with model usage then billed at API rates. In exchange, organisations get the compliance stack — audit logs, custom data retention, access management and support arrangements. Two organisations rarely pay the same Enterprise bill, because so much of it is metered usage.
The hidden variable: how limits actually work
The number that decides whether a plan fits is not the price but the limit, and Claude’s limits are measured in usage, not messages. A short question costs little; a long document analysis or an hour of Claude Code costs a great deal. Two users on the same plan can have wildly different experiences of “enough.”
Anthropic meters usage in rolling windows — the Max tiers are literally described as multiples of Pro’s usage per session window — with additional weekly caps on the heaviest features. Practical consequence: bursty users hit walls that steady users never see, and agent work consumes allowances at a pace chat never does.
Also keep the boundary clear: subscriptions do not include general API credit. Developers pay for API tokens separately, at per-model rates, and businesses often end up buying both — seats for staff, tokens for products. Budgeting for Claude means budgeting for two products that happen to share a name.
Finally, prices and structures on this page were read from Anthropic’s pricing page in October 2026. This industry reprices constantly; the tiers above are a reliable map of the logic, and the live page is the authority for the numbers.
Which plan is actually for you?
Occasional questions and light drafting: stay free. Daily professional use: Pro, and it is not close. All-day coding agents: Max, sized by how often you hit Pro’s wall in a trial month. A company rolling Claude out to staff: Team, with Premium seats reserved for the engineering minority who will genuinely use them.
One budgeting warning applies to every tier: limits are measured in usage, not messages, and long documents and agent runs consume far more than short chats. The right plan is the one whose ceiling you hit rarely — not the one whose brochure sounds best.
0 Comments