Claude Pricing 2026: Every Plan, API Rate, and How to Pay Less

Claude (claude.ai) is Anthropic's AI assistant — used by individuals for writing and research, by developers building applications via the API, and by enterprises deploying AI at scale. Pricing works two ways: subscription plans for claude.ai users, and token-based API pricing for developers.

This guide covers every plan tier, every current API rate, every discount mechanism, and real cost examples so you know exactly what you will pay before signing up.

---

Claude Subscription Plans (claude.ai)

Free

$0/month — no credit card required

The free plan gives access to Claude on web, iOS, Android, and desktop. Includes text generation, image understanding, code generation, web search, and desktop extensions. The limit is a daily usage cap — you will hit it if you use Claude heavily throughout the day, but it is genuinely useful for occasional use.

Includes:

• Claude Sonnet (with daily limits)
• Web, iOS, Android, desktop access
• File uploads and image analysis
• Web search
• Desktop extensions

---

Pro

$20/month

Pro removes the daily cap and adds Claude Code — Anthropic's agentic coding tool for terminal use. Key additions over free:

• Claude Code in terminal (agentic coding)
• File creation and code execution
• Unlimited Projects
• Google Workspace integration
• Remote MCP connectors
• Extended thinking (reasoning) models
• Higher usage limits than free

For developers and power users who use Claude daily, Pro is the core tier.

---

Max

$100/month (5× Pro usage) or $200/month (20× Pro usage)

Max is for users who hit Pro's limits — developers running long agentic sessions, researchers doing intensive multi-document analysis, or teams using Claude as a core tool all day. The 20× tier ($200/mo) is the highest individual subscription:

• Priority access to new features and models
• ~800 prompts in a 5-hour window at Max 20×
• Priority queue during peak hours

---

Claude Team Plans

Team plans require a minimum of 5 seats.

Critical note: Team Standard does not include Claude Code. Claude Code only ships with Team Premium at $150/seat/month.

---

Claude Enterprise

Custom pricing — contact Anthropic sales.

Enterprise adds everything in Team plans plus:

• Expanded context window
• Role-based access control
• SCIM provisioning
• Audit logging
• Compliance API
• Custom data retention policies
• Google Docs catalog integration

---

Claude API Pricing

API pricing is per million tokens (MTok). A token is roughly 0.75 English words.

Current Models — Recommended

Opus 4.7 tokenizer note: The new tokenizer produces up to 35% more tokens per request compared to Opus 4.6 for some content types. Factor this into cost estimates when migrating from 4.6 to 4.7.

Legacy Models — Still Available

Legacy Opus 4.1 at $15/$75 per MTok is dramatically more expensive than current Opus 4.6/4.7 at $5/$25. Migrating from Opus 4.1 to Opus 4.6 saves 67% immediately.

---

Additional API Features and Costs

---

Cost Reduction: Prompt Caching

Prompt caching stores repeated context so subsequent requests pay a fraction of the standard input rate.

Example: A Sonnet 4.6 application with a 50,000-token system prompt sent 10,000 times/day:

• Without caching: 50,000 × 10,000 × $3.00/M = $1,500/day
• With caching: First request (full) + 9,999 × $0.30/M × 50,000 = $150/day
• Saving: 90%, or ~$40,500/month

Prompt caching is the single highest-leverage cost reduction available on the Claude API.

---

Cost Reduction: Batch API

The Batch API processes requests asynchronously — results returned within 24 hours — at 50% off all standard token rates.

Batch API rates:

Best for: Content processing pipelines, document analysis, evaluation datasets, embedding generation, any workload that does not require real-time results.

Combined savings: Batch API + prompt caching together can reduce costs by up to 95% compared to standard on-demand pricing for eligible workloads.

---

Model Selection: Which Claude to Use

The right model choice is the most important cost decision — not prompt length optimisation.

Use Haiku 4.5 ($1/$5) for:

• Classification tasks (is this message a complaint or a question?)
• Simple data extraction (pull the date from this invoice)
• High-volume routing or preprocessing
• Any task where speed and cost matter more than reasoning depth

Use Sonnet 4.6 ($3/$15) for:

• Production applications requiring strong reasoning
• Code generation and debugging
• Long-document summarisation and analysis
• The majority of customer-facing AI features

Use Opus 4.7 ($5/$25) for:

• Complex multi-step reasoning
• Advanced code architecture and review
• Tasks where Sonnet's output is not good enough after testing
• Research and analysis requiring frontier reasoning

The key principle: Default to Sonnet. Escalate to Opus only when Sonnet provably falls short. Route simple tasks to Haiku. This model-routing approach typically reduces API costs by 60–80% versus using Opus for everything.

---

Real Cost Examples

Startup customer support — Sonnet 4.6 with caching

5M input tokens + 2M output tokens per month, with 80% of input tokens cached:

• Cached input: 4M × $0.30 = $1.20
• Uncached input: 1M × $3.00 = $3.00
• Output: 2M × $15.00 = $30.00
• Monthly total: ~$34–50/month

Enterprise migration — Opus 4.1 to Opus 4.6

Same 100M tokens/month workload:

• Opus 4.1: $15 × 100M / 1M = $1,500/month
• Opus 4.6: $5 × 100M / 1M = $500/month
• Saving: $1,000/month (67%)

High-volume content — Haiku 4.5 + Batch API

20M input + 10M output tokens:

• Standard Haiku: (20 × $1) + (10 × $5) = $70/month
• Batch Haiku: (20 × $0.50) + (10 × $2.50) = $35/month

Developer team with code execution

1M input + 500K output tokens + 1,500 paid container hours:

• Tokens: $3 + $7.50 = $10.50
• Code execution: 1,500 × $0.05 = $75
• Total: ~$85.50/month

---

Claude vs ChatGPT vs Gemini API Pricing

Claude Sonnet 4.6 is priced competitively with GPT-4o (openai.com) and above Gemini 2.5 Pro (gemini.google.com). Claude's primary advantage is writing quality and long-context performance — not raw price. For cost-sensitive, high-volume workloads, GPT-4o mini or Gemini Flash are cheaper.

---

Frequently Asked Questions

How much does Claude cost per month? The free plan is $0. Claude Pro is $20/month. Claude Max is $100/month (5× usage) or $200/month (20× usage). Team Standard is $25/user/month (annual). Team Premium is $150/user/month.

What is the cheapest Claude API model? Claude Haiku 4.5 at $1.00/$5.00 per million tokens is the most affordable current model. Legacy Haiku 3 at $0.25/$1.25 is cheaper but represents older technology. Haiku 4.5 is the recommended budget option for production use.

Does the free plan include Claude Opus? No. Free and Pro plans include Claude Sonnet. Claude Opus 4.7 is accessible on Max plans and via API (paid).

What is the Batch API discount? The Batch API processes requests within 24 hours at 50% off all standard token rates. It applies to all current Claude models and is the easiest cost reduction for non-real-time workloads.

Is Claude API or subscription better for developers? Depends on usage intensity. If you use Claude Code 3+ days per week with Opus-heavy tasks, Max 20× ($200/mo) often wins economically over pay-per-token. For lighter or Sonnet-focused use, API pay-as-you-go is typically cheaper. A hybrid (subscription + API overflow) is practical for teams.

How does prompt caching work? Frequently repeated context (system prompts, knowledge bases, conversation history) is stored server-side. Subsequent requests that use the cached content are charged at 10% of the standard input rate — a 90% discount on those tokens.

---

Building with Claude? List your AI product on Startup Launch Page and reach investors and early adopters actively looking for new tools.