How Much Does Claude Cost?

Complete pricing guide for every Claude plan and API model. Updated for 2026.

Claude Plans at a Glance

Choose the plan that fits your usage. All prices in USD.

Free
$0/mo

Try Claude with no commitment. Basic access for light use.

  • +Access to Claude Sonnet
  • +Limited daily messages
  • +Text generation, analysis, code help
  • -No extended thinking
  • -No premium models
Most Popular
Pro
$20/mo

For professionals who use Claude daily.

  • +Claude Sonnet 4 + Haiku priority
  • +5x more usage than Free
  • +Extended thinking
  • +Claude Code (with limits)
  • +Projects and file uploads
Max
$100 / $200/mo

Power users and heavy Claude Code users.

  • +$100: 5x Pro limits, some Opus
  • +$200: 20x Pro limits, full Opus 4
  • +Extended context windows
  • +Highest priority access
  • +Unlimited Claude Code
Team
$25/user/mo

For teams of 2-50 who need managed access.

  • +Everything in Pro per user
  • +Admin console and permissions
  • +Shared projects and knowledge bases
  • +Higher per-user limits than Pro
  • +Data not used for training
Enterprise
Custom pricing

For large organisations with compliance and security requirements.

  • +SSO (SAML) and SCIM provisioning
  • +Custom data retention policies
  • +Dedicated account management
  • +SLA guarantees and audit logs
  • +Advanced admin analytics
  • +HIPAA and SOC 2 compliance

Feature Comparison

Side-by-side breakdown of what each plan includes.

FeatureFreeProMaxTeamEnterprise
Monthly price$0$20$100/$200$25/userCustom
Model accessSonnetSonnet 4, HaikuAll incl. Opus 4Sonnet 4, HaikuAll models
Message limitsBasic5x Free5-20x ProHigher than ProCustom
Context windowStandardStandardExtendedStandardExtended
Extended thinkingNoYesYesYesYes
File uploadsNoYesYesYesYes
Claude CodeNoLimitedUnlimitedLimitedCustom
ProjectsNoYesYesSharedShared
ArtifactsBasicYesYesYesYes
Priority accessNoneStandardHighestStandardDedicated
Admin consoleNoNoNoYesAdvanced
SSO / SCIMNoNoNoNoYes
Data training opt-outNoNoNoYesYes

API Pricing by Model

Pay per token with no minimum commitment. All prices per million tokens.

ModelInput / 1M tokensOutput / 1M tokensContext
Claude Opus 4$15.00$75.00200K
Claude Sonnet 4$3.00$15.00200K
Claude Haiku 3.5$0.80$4.00200K

Pricing is approximate and may change. Check anthropic.com/pricing for the latest rates.

API Cost Savings

Two built-in features can dramatically reduce your API costs.

Prompt Caching (up to 90% off)

Cache frequently used system prompts and static context. Subsequent requests read from cache at a fraction of the cost.

Standard input tokensBase rate
Cache write (first request)+25% vs base
Cache read (subsequent)-90% vs base
Cache TTL5 minutes (refreshes on use)

Best for: apps with repeated system prompts, RAG with static context, multi-turn conversations.

Batch API (50% off)

Submit jobs asynchronously and receive results within 24 hours. All token costs are halved.

Input token discount50% off
Output token discount50% off
Turnaround timeUp to 24 hours
Real-time use?No (async only)

Best for: data processing, evaluation, content generation at scale.

Prompt Caching: How It Works

Prompt caching stores frequently used content (like system prompts or large documents) on Anthropic's servers so that subsequent API calls can reference it without re-sending and re-processing those tokens.

When to use it: Any time your requests include repeated preamble text. A chatbot with a 2,000-token system prompt making 1,000 calls per day would process 2 million input tokens just for the system prompt. With caching, the first call writes to cache (costing 1.25x), and the remaining 999 calls read from cache (costing 0.1x each). That turns 2M tokens at full price into roughly 200K equivalent tokens.

Cost Savings Example (Sonnet 4)

ScenarioWithout CachingWith CachingSavings
1K requests/day, 2K system prompt$180/mo$21/mo88%
500 requests/day, 8K RAG context$360/mo$48/mo87%

Tip: Cache entries expire after 5 minutes of inactivity. If your request volume is steady, the cache stays warm automatically. For bursty workloads, consider batching requests to keep the cache alive.

Real-World API Cost Examples

What typical production workloads cost per month.

Use CaseModelMonthly VolumeEstimated Cost
Customer support chatbot (1K conversations/day)Sonnet 4~60M tokens$200-$400/mo
Document summarization (500 docs/day)Haiku 3.5~30M tokens$30-$60/mo
Code review tool (200 PRs/week)Sonnet 4~40M tokens$120-$250/mo
Research assistant (complex, multi-step)Opus 4~10M tokens$200-$500/mo
Content moderation (10K items/day)Haiku 3.5~100M tokens$80-$150/mo
RAG application (2K queries/day)Sonnet 4~50M tokens$150-$300/mo

API Cost Estimator

Estimate your monthly Claude API costs based on usage patterns.

1010,000

Monthly requests: 3,000

Input tokens: 6.0M

Output tokens: 2.4M

Claude Opus 4

$270.00/month
Input: $15/1M tokensOutput: $75/1M tokens

Claude Sonnet 4

$54.00/month
Input: $3/1M tokensOutput: $15/1M tokens

Claude Haiku 3.5

Best Value
$14.40/month
Input: $0.8/1M tokensOutput: $4/1M tokens

Same workload on other providers

GPT-4o$39.00/mo
Gemini 1.5 Pro$19.50/mo

Recommendation: For your chatbot / customer support workload, Claude Haiku 3.5 offers the best balance of quality and cost at $14.40/month. Enable prompt caching or the Batch API to reduce costs further.

Claude vs ChatGPT vs Gemini

How Claude compares on pricing, models, and capabilities.

Consumer Plans

Claude ProChatGPT PlusGemini Advanced
Price$20/mo$20/mo$20/mo
Best model includedSonnet 4 + some OpusGPT-4o + some o1Gemini 1.5 Pro
Usage limitsModerateModerateGenerous
Context window200K tokens128K tokens1M tokens
Key strengthExtended thinking, codeImage gen, pluginsGoogle integration

API Pricing: Flagship Models (per 1M tokens)

Claude Sonnet 4GPT-4oGemini 1.5 Pro
Input$3.00$2.50$1.25
Output$15.00$10.00$5.00

API Pricing: Budget Models (per 1M tokens)

Claude Haiku 3.5GPT-4o miniGemini 1.5 Flash
Input$0.80$0.15$0.075
Output$4.00$0.60$0.30

Honest assessment: Claude is not the cheapest API option. Its value comes from reasoning quality, safety, and long-context performance. For pure cost optimisation, Gemini and GPT-4o mini are cheaper per token. For quality-sensitive tasks where accuracy matters (complex reasoning, nuanced writing, code generation), Claude Opus and Sonnet are competitive. Choose based on your quality requirements, not just price.

Which Claude Plan Is Right for You?

A simple decision guide based on how you use Claude.

Student or hobbyist

Fewer than 10 messages per day, exploring what Claude can do

Free tier

Daily professional user

10-50 messages per day, using Claude for work tasks

Pro ($20/mo)

Power user or developer

Heavy Claude Code usage, need Opus 4, high daily volume

Max ($100 or $200/mo)

Small to mid-size team

5-50 team members, need admin controls and shared projects

Team ($25/user/mo)

Enterprise with compliance needs

50+ users, SSO, audit logs, HIPAA or SOC 2 requirements

Enterprise (contact sales)

Building an application

Integrating Claude into your product via the API

API (start with Haiku for prototyping)

High-volume production workload

Thousands of requests daily, cost is a priority

API + Batch + Caching (50-90% savings)

Frequently Asked Questions

How much does Claude Pro cost?
Claude Pro costs $20 per month. It gives you priority access to Claude Sonnet 4 and Haiku, 5x more usage than the free tier, extended thinking, Claude Code access, projects, and file uploads.
Is there a free version of Claude?
Yes. Claude offers a free tier with access to Claude Sonnet, a limited daily message allowance, and basic features including text generation, analysis, and code help. No credit card required.
How much does the Claude API cost?
Claude API pricing varies by model. Claude Opus 4 costs $15 input / $75 output per million tokens. Sonnet 4 costs $3 input / $15 output. Haiku 3.5 costs $0.80 input / $4 output. You can reduce costs by up to 90% with prompt caching or 50% with the Batch API.
What is the difference between Claude Pro and Max?
Claude Pro costs $20/month and gives 5x the free tier usage. Claude Max has two tiers: $100/month (5x Pro usage plus some Opus access) and $200/month (20x Pro usage, full Opus 4 access, and highest priority). Max is designed for power users and heavy Claude Code users.
Can I set a spending limit on the Claude API?
Yes. You can set a monthly spending cap through the Anthropic dashboard. Once you hit your limit, API requests will be rejected until the next billing cycle or until you raise your cap.
What are tokens and how do they relate to cost?
Tokens are the units Claude uses to process text. Roughly 1 token equals 0.75 words, or about 4 characters. You pay per token for both your input (what you send) and output (what Claude generates). System prompts also count as input tokens.
How does prompt caching save money?
Prompt caching lets you cache frequently used system prompts and context. Cached input tokens cost 90% less than standard input tokens. The initial write to cache costs 25% more, but subsequent reads are dramatically cheaper. Cache entries last 5 minutes and refresh on use.
Is Claude cheaper than ChatGPT?
At the consumer level, Claude Pro and ChatGPT Plus both cost $20/month. At the API level, Claude Sonnet 4 ($3/$15 per 1M tokens) is more expensive than GPT-4o ($2.50/$10). For budget models, Claude Haiku 3.5 ($0.80/$4) costs more than GPT-4o mini ($0.15/$0.60). Claude competes on reasoning quality and long-context performance rather than being the cheapest option.