Claude AI pricing: every plan, product, and API rate

Anthropic sells Claude as a consumer subscription, a team seat, and a metered API, and the same model costs wildly different amounts depending on which door you came through. Here is all of it, with the multipliers that most price lists leave out.

By the Continuum team. We build a workbench that runs Claude Code, Codex, and their peers, so the model rates quoted here are the ones our own cost analytics ship with.

The short version

As of August 2026 Claude costs $0 on Free, $20/mo on Pro ($17/mo billed annually at $200 up front), $100/mo for Max 5x, and $200/mo for Max 20x. Team seats are $20/seat/mo billed annually ($25 monthly) for a standard seat and $100/seat/mo annually ($125 monthly) for a premium seat with 5x the usage. Enterprise is a seat price plus usage at API rates. The API bills per million tokens: Opus 5 at $5 in and $25 out, Sonnet 5 at $2 and $10 through 31 August 2026, then $3 and $15, Haiku 4.5 at $1 and $5. Claude Code is included from Pro upward.

What you need to know
  • Pro is $20/mo, or $17/mo billed annually as $200 up front. It includes Claude Code.
  • Max is $100 (5x) or $200 (20x). Same models, same features, more session capacity.
  • Team seats are $20 or $100 per seat per month billed annually. The premium seat is 5x the standard seat.
  • The API is a separate billing system: no subscription, no windows, billed per token.
  • Sonnet 5 is on introductory pricing of $2 in and $10 out per million through 31 August 2026, then $3 and $15.
  • Cache reads are 0.1x base input, a 5-minute cache write is 1.25x, and a 1-hour cache write is 2x.

Consumer and team plans

Claude subscription plans, read from Anthropic in August 2026.

PlanPriceClaude CodeWhat it is for
Free$0NoChat and code generation only. Claude Code requires Pro or higher.
Pro$20/mo, or $17/mo annually ($200 up front)YesIndividual use, part-time agent work.
Max 5x$100/moYes5x Pro capacity per session. A full working day of agent coding.
Max 20x$200/moYes20x Pro capacity per session. All-day, parallel, Opus-heavy work.
Team standard seat$20/seat/mo annually, $25 monthlyYesShared billing, admin controls, more usage than Pro.
Team premium seat$100/seat/mo annually, $125 monthlyYes5x standard-seat usage for the people who live in the agent.
EnterpriseSeat price plus usage at API ratesYesSSO, audit, procurement, analytics API.
APIPer tokenYes, with a keyProgrammatic use, CI, spiky workloads.

What the multipliers actually mean

Anthropic describes Max 5x as "5 times more usage per session than the Pro plan" and Max 20x as "20 times more usage per session". Note the words per session. The multiplier applies to the rolling five-hour session allowance. The weekly caps that sit above it are their own numbers and do not scale by the same factor, which is why a Max 20x account can still meet a weekly ceiling in a heavy week.

API rates

The API has no subscription and no usage windows. You are billed per million tokens, separately for what you send, what you cache, and what you get back.

Per-million token rates, read from Anthropic in August 2026. MTok = million tokens.

ModelInput5m cache write1h cache writeCache readOutput
Claude Opus 5$5.00$6.25$10.00$0.50$25.00
Claude Opus 4.8$5.00$6.25$10.00$0.50$25.00
Claude Sonnet 5 (from 1 Sep 2026)$3.00$3.75$6.00$0.30$15.00
Claude Sonnet 5 (through 31 Aug 2026)$2.00$2.50$4.00$0.20$10.00
Claude Sonnet 4.6$3.00$3.75$6.00$0.30$15.00
Claude Haiku 4.5$1.00$1.25$2.00$0.10$5.00

Caching changes the effective rate

Prompt caching lets a stable prefix be written once and re-read cheaply on later turns. The multipliers against base input are fixed and published: a 5-minute cache write is 1.25x, a 1-hour cache write is 2x, and a cache read is 0.1x. So a 5-minute cache pays for itself after one read, and a 1-hour cache after two.

For an agent session, where the same system prompt and file context are re-sent every turn, this is the difference between an affordable session and an alarming one.

The modifiers most price lists omit

Multipliers and add-ons that stack on top of the base rates.

ModifierEffectWhen it applies
Batch API50 percent off input and outputAsynchronous jobs that tolerate latency
Data residency1.1x on every token categoryClaude 4.6+ with inference_geo: "us"
Fast mode$10 in / $50 out per MTokOpus 5 and Opus 4.8, research preview, API only
Web search tool$10 per 1,000 searchesPlus the tokens the results consume
Web fetch toolNo extra chargeYou still pay for the fetched content as tokens
Code execution$0.05 per container-hour after 1,550 free hours/moFree when used with web search or web fetch
Long contextNo premiumClaude 4.6+ include the full 1M window at standard rates

Which door to come through

The same work costs very different amounts depending on the billing system, and the decision is about the shape of your usage rather than its size.

Your patternTake
You use Claude chat daily and code with it sometimesPro. The chat alone justifies it.
You code with an agent most working daysMax 5x. Anthropic's own metered average is $150 to $250 a month.
You run agents all day, in parallel, on the big modelMax 20x.
You are five engineers who all want itTeam standard seats, premium for the two heaviest.
You are building a product on top of ClaudeAPI. Subscriptions are not licensed for that shape.
You run agents in CI or on a scheduleAPI. A subscription is for interactive use.
Your usage is a heavy month then nothingAPI, or buy one month of Max and cancel.

Nothing stops you having both, and plenty of people do: a subscription for interactive work and a key for scripts. They bill independently and do not draw from each other.

Where the spend actually shows up

Every billing system reports somewhere different, and the surface you look at determines whether you can attribute a number to a person or a project at all.

Where each billing system publishes what you spent.

How you payWhere you see spendWhere you cap itPer-person detail
Pro or MaxSettings then Usage; usage bars, not dollarsUsage-credit monthly capn/a, it is one person
Team or Enterprise seatOrg spend report, per user and per model, CSV, dailyUsage-credit spend limits per org, group, or memberYes, in the spend report
Claude Console API keyConsole usage page and the Claude Code dashboardWorkspace spend limitsYes, via the Claude Code Analytics API
Enterprise planSame as Team, plus the Enterprise Analytics APISame as TeamYes, programmatically
Bedrock, Google Cloud, Microsoft FoundryYour cloud billing consoleYour cloud budget controlsOnly via OpenTelemetry or a gateway

Discounts, credits, and edge cases

  • Annual billing on Pro is $200 up front, roughly $17/mo, about 15 percent off the monthly price.
  • Annual billing on Team takes a standard seat from $25 to $20 and a premium seat from $125 to $100, so 20 percent off both.
  • Usage credits let Pro and Max subscribers keep working past the included limit, billed at standard API rates, with a monthly spending cap, optional auto-reload, and a $2,000 daily redemption limit. Team and Enterprise admins turn them on and set limits per organization, group, or member.
  • Batch API runs asynchronous jobs at 50 percent off both input and output. It stacks with caching.
  • Volume and academic discounts are negotiated case by case through Anthropic sales rather than published.
  • Marketplace billing (Claude Platform on AWS, Claude in Microsoft Foundry) invoices in Claude Consumption Units at $0.01 per CCU, rating your tokens at the same standard per-model rates first.

Questions people ask

$20 a month, or $17 a month if you pay annually, billed as $200 up front. It includes Claude Code.

There is a Free plan at $0 with limited usage, but it does not include Claude Code. Anthropic's setup docs state that Claude Code requires a Pro, Max, Team, Enterprise, or Console account, as of August 2026.

Session capacity only. Anthropic describes them as 5 times and 20 times more usage per session than Pro, at $100 and $200 a month. Same models, same features, same Claude Code.

Per million tokens: Opus 5 is $5 input and $25 output, Sonnet 5 is $3 and $15 standard ($2 and $10 on introductory pricing through 31 August 2026), and Haiku 4.5 is $1 and $5. Cache reads are a tenth of base input.

No. Subscriptions and the API are separate billing systems. An API key is created and billed independently in the Claude Console. The one connection is usage credits, which bill subscription overage at API rates.

A standard seat is $20 per seat per month billed annually, or $25 monthly. A premium seat is $100 per seat per month annually, or $125 monthly, and carries 5x the standard seat usage. Enterprise adds a seat price plus usage at API rates.

No. Subscriptions are per person, and usage windows are per account, so two people on one login halve each other constantly. Team plans exist for this and add central billing, provisioning, and clean offboarding.

No. Claude 4.6 and later models include the full 1 million token context window at standard pricing, so a 900k-token request is billed at the same per-token rate as a 9k one. The relevant change now is the newer tokenizer on 4.7-generation models, which produces about 30 percent more tokens for the same text.

Sources

Every figure above was read from these pages on August 2026. Vendors reprice without notice; if you find a stale number, tell us.

  1. Anthropic plans and pricing
  2. Anthropic API pricing
  3. Anthropic help: usage limits
Try it

One gauge for
every subscription.

Continuum shows live five-hour and weekly headroom for every Claude and Codex account you run, plus what each one has cost, in the menu bar and on your phone.

free app · your subscriptions · local-first