Anthropic Claude licensing

Anthropic Claude licensing is how Anthropic sells access to its Claude models to organizations and individuals. Business and developer use falls under the Commercial Terms of Service. Those terms govern “Customer’s use of Anthropic API keys and any other Anthropic offerings that references these Terms” and are “not for consumer use”.^[1] Individuals use Claude.ai and Claude Pro under the Consumer Terms of Service.^[13] Two metric families apply: per-user seats (with optional usage credits) for Claude plans, and per-token consumption for the Claude API. Enterprise combines both: “a fixed seat fee and separate usage charges.”^[7] The article describes the position as published on 2026-09-26. It follows the general model in Subscription and consumption licensing.

Plans

Team seats are Standard or Premium, and an organization can mix them. Standard seats include “1.25x the Pro plan’s per-session usage allowance” and Premium seats “6.25x”. Both have a weekly limit across all models.^[5] All new Enterprise contracts use a single Enterprise seat. Legacy Chat, Chat + Claude Code, and Standard/Premium Enterprise seat types move to it at renewal.^[7]

Metrics

The catalog records nine Anthropic metrics under Anthropic’s own names. It gives no conversion between seats and tokens because Anthropic publishes none, other than the rule that Enterprise usage and plan overage run “at standard API rates”.^[2]

For scale, a few flagship list prices per MTok, as of 2026-09-26:

• Claude Opus 5.5: $4 base input, $20 output
• Claude Sonnet 5: $2 base input, $10 output
• Claude Haiku 4.5: $1 base input, $5 output

For Sonnet 5, the pricing page says the $2/$10 price announced at launch as introductory “is now the standard price”.^[3] The full model table is on the pricing page. The catalog does not copy it.

Counting and metering

Team plan seats

A Team plan has “a minimum of two members” and supports up to 150 seats (rule Team plan two-member minimum and 150-seat maximum, dated 2026-09-26).^[5] The billed quantity is members: “You will be charged for the number of members on your team at the beginning of each billing cycle” (Team plan billed on members at cycle start).^[6]

• Adding members or upgrading seats is prorated and charged immediately.
• Removing a member gives no immediate credit. The seat becomes available to someone else.^[6]
• Within the purchased allocation, admins can move users between Standard and Premium.
• Reducing the seat count, or moving a member from Premium to Standard, takes effect at the next renewal with no prorated refund.
• Members set to “No seat assigned” do not count toward the seat limit, but pending invitations do (Team plan seat reassignment and reductions).^[4]

Usage limits apply to each member, not to the team as a whole.^[5] The Named user licensing article covers this general pattern.

Enterprise seats and usage

Enterprise seats are charged at the start of the annual billing cycle, and added seats are prorated immediately. “Seats cannot be removed mid-term on self-serve Enterprise plans.” “Enterprise seats don’t come with an individual token allowance. All usage across your organization is billed together at API rates” (Enterprise seat fee plus usage at API rates).^[7]

• Self-serve: usage is bought upfront in shared credits, and use stops when they run out.
• Sales-assisted: usage is invoiced monthly in arrears.
• US-only inference: when enabled, usage is billed at 1.1x for Claude Opus 4.6, Claude Sonnet 4.6 and later models.^[7]

Claude API

Commercial Terms §H.1 makes the customer responsible for fees “at the rates specified on the Model Pricing Page, unless otherwise agreed by the parties”. Rate updates take effect at the earlier of 30 days after posting or Notice to the customer (API fees at Model Pricing Page rates, effective 2025-06-17).^[1]

• Prompt caching: a 5-minute cache write costs 1.25x the base input price, a 1-hour cache write 2x, and a cache read 0.1x. Two exceptions: 0.025x on Claude Fable 5.1 and Claude Mythos 5.1, and 0.05x on Claude Opus 5.5 (Prompt caching multipliers).^[3]
• Batch API: “a 50% discount on both input and output tokens”. It can be combined with prompt caching (Batch API 50% discount).^[3]
• Data residency: US-only inference applies “a 1.1x multiplier on all token pricing categories” for Claude 4.6 and later models (US-only inference 1.1x multiplier).^[3]
• Rate limits: set at the organization level and measured in RPM, ITPM and OTPM. Organizations move between tiers automatically. Monthly spend caps are $500 (Start), $1,000 (Build) and $200,000 (Scale), and the Custom tier has no cap (API usage tiers and monthly spend caps).^[11]
• Cache-aware ITPM: “For most Claude models, only uncached input tokens count toward your ITPM rate limits” (Cache-aware ITPM).^[11]

Priority Tier was Anthropic’s reserved-capacity offer. “Priority Tier capacity commitments are no longer available for purchase.” Existing commitments run to their contract end date (Priority Tier no longer purchasable).^[12] Those commitments specify input and output tokens per minute, a duration of 1, 3, 6 or 12 months, and one model version.^[12]

Data use and retention

Commercial Terms §B: the customer retains all rights to Inputs and owns Outputs, and “Anthropic may not train models on Customer Content from Services” (No training on Customer Content, effective 2025-06-17).^[1] The pricing page lists “No model training on your content by default” for the Team plan.^[2]

• API: “For Anthropic API users, we automatically delete inputs and outputs on our backend within 30 days of receipt or generation”. Exceptions cover services with retention under customer control, agreed arrangements such as zero data retention, Usage Policy enforcement, and legal requirements. Chats flagged for Usage Policy violations are kept up to 2 years, with classification scores up to 7 years (API inputs and outputs deleted within 30 days).^[8]
• Claude for Work products: chats are kept in the product. When a user deletes a conversation, it is removed from back-end storage within 30 days.^[8]
• Enterprise custom retention: “The minimum retention period is 30 days”. Without a custom setting, data is retained indefinitely (Enterprise custom retention minimum 30 days).^[9]
• Individual plans: Anthropic may use Materials to develop products “including training our models, unless you opt out of training through your account settings” (Consumer Materials used for training unless opted out).^[13]

Usage restrictions

Commercial Terms §D.4 bars using the Services “to build a competing product or service, including to train competing AI models or resell the Services except as expressly approved by Anthropic”. It also bars reverse engineering or duplicating the Services (Commercial use restrictions).^[1] The Usage Policy (effective 2025-09-15) applies to anyone who can submit inputs, “including via any authorized resellers or passthrough access”. Violations may lead Anthropic to “throttle, suspend, or terminate” access (Usage Policy applies to all users).^[10] The Consumer Terms forbid sharing account login information or API keys.^[13] For general compliance method, see License compliance.

Cloud and marketplace availability

Anthropic states that Claude models are available on Amazon Bedrock and Google Cloud, “where the cloud provider invoices you”. Official pricing for those platforms is on the cloud providers’ own pages.^[3] For Claude Sonnet 4.5, Haiku 4.5, Opus 4.5 and later models, “Regional and multi-region endpoints include a 10% premium over global endpoints” (Partner cloud regional endpoint premium).^[3]

• Amazon Bedrock: Claude “runs on AWS-managed infrastructure with zero operator access”. Data handling is governed by Amazon Bedrock. The default quota is 2 million input tokens per minute.^[15]
• Google Cloud: Anthropic describes availability through Google Cloud’s Agent Platform. It offers global, multi-region and regional endpoints. Data handling is governed by Google Cloud.^[16]
• Claude Platform on AWS and Claude in Microsoft Foundry: these are Anthropic-operated and bill through AWS Marketplace or the Azure Marketplace in Claude Consumption Units. Anthropic rates usage at standard API rates, applies discounts, and converts at “$0.01 per CCU”. Metering is hourly and invoicing monthly in arrears (Marketplace usage invoiced in CCU).^[3]
• Enterprise via AWS Marketplace: the pricing page lists “AWS Marketplace availability” for sales-assisted Enterprise plans.^[2]

Programs

Out of scope

This article does not cover:

• Claude Managed Agents session-runtime pricing
• Fast mode rates
• Web search, code execution and other tool prices
• Claude Mythos models, which are offered by invitation under Project Glasswing
• The Data Processing Addendum, Service Specific Terms and Supplemental Credits Terms, which were not retrieved
• Claude for Government
• US K-12 terms

Consumer Terms were retrieved in the version for EEA and Switzerland residents. The version for other regions was not served on retrieval.^[13] For the general nature of a software license, see that article.

Catalog rows

Sample catalog rows for this article, grouped by table. All rows are on the Anthropic vendor page.

• Metrics
  • Standard seat
  • Premium seat
  • Enterprise seat
  • Base input tokens
  • Output tokens
  • Claude Consumption Unit
• Rules
  • Team plan two-member minimum and 150-seat maximum
  • Team plan billed on members at cycle start
  • Enterprise seat fee plus usage at API rates
  • No training on Customer Content
  • API inputs and outputs deleted within 30 days
  • API usage tiers and monthly spend caps
  • Prompt caching multipliers
• Programs
  • Claude Team plan
  • Claude Enterprise plan
  • Claude Platform on AWS and Claude in Microsoft Foundry