This site is independently operated and is not affiliated with Anthropic. Verify pricing on Anthropic's official website.

Updated March 2026

Claude API Pricing in 2026

Anthropic offers three Claude models ranging from $0.80 to $15 per million input tokens. This guide breaks down every cost, explains caching and batch discounts, and helps you choose the right model for your workload.

Haiku 3.5

$0.80/MTok

Popular

Sonnet 4

$3.00/MTok

Opus 4

$15.00/MTok

Calculate Your Monthly Cost →

Complete Claude API Pricing Table

All prices are per 1 million tokens. Prompt caching write costs 25% more; cache reads cost 90% less. Batch API is 50% off standard prices.

ModelInputOutput
Claude Opus 4$15.00$75.00
Claude Sonnet 4Most Popular$3.00$15.00
Claude Haiku 3.5$0.80$4.00

What does “per million tokens” mean?

1M tokens≈ 750,000 words
1M tokens≈ 1,500 pages of text
1 token≈ 4 characters of English

How Claude API Pricing Works

The Claude API uses token-based pricing. Every piece of text you send to the API (your prompt, system instructions, conversation history) is split into tokens—small chunks of text that the model processes. You pay separately for input tokens (what you send) and output tokens (what Claude generates back).

Output tokens are always more expensive than input tokens because generating text requires more computation than reading it. On Claude Sonnet 4, output costs 5x more than input ($15 vs $3 per million tokens). This means a long prompt with a short response is cheaper than a short prompt with a long response.

Prices are quoted per million tokens (MTok). To put that in perspective: one million tokens equals roughly 750,000 words, which is about 1,500 pages of standard text or roughly 10-15 full-length novels. A single customer support interaction typically uses 500-2,000 input tokens and 200-1,000 output tokens—a tiny fraction of a million.

The cost formula

Cost = (input_tokens / 1,000,000 × input_price) + (output_tokens / 1,000,000 × output_price)

Example: 2,000 input tokens + 500 output tokens on Sonnet 4 = (2,000 / 1M × $3) + (500 / 1M × $15) = $0.0135 per request.

There is no monthly subscription, no minimum spend, and no commitment. You pay only for the tokens you consume. Anthropic bills based on actual usage, making it easy to start small and scale as your application grows. You can also reduce costs significantly through prompt caching (save up to 90% on input) and the Batch API (50% off everything).

Quick Cost Examples

What common API requests actually cost, calculated with real pricing.

Chatbot conversation

500 input + 200 output tokens on Sonnet 4

<$0.01

per request

Summarise a 50-page document

~30,000 input + 1,000 output tokens on Sonnet 4

$0.10

per request

10,000 support tickets via Batch API

2,000 input + 500 output each on Haiku Batch (50% off)

$18.00

per request

The batch example shows the total cost for all 10,000 tickets combined ($$18.00 total), not per request.

Choose the Right Claude Model

Each model sits at a different point on the cost-quality spectrum. Pick the cheapest one that meets your quality bar.

Most Popular

Claude Sonnet 4

$3 / $15 input / output per MTok

The sweet spot. Handles complex coding, analysis, and content generation with quality close to Opus at a fraction of the price.

Best for:

  • Production chatbots and assistants
  • Code generation and review
  • Content creation at scale
  • RAG-powered Q&A systems
Full pricing breakdown →
Premium

Claude Opus 4

$15 / $75 input / output per MTok

Maximum intelligence for tasks where quality matters more than cost. Excels at multi-step reasoning, research synthesis, and architectural decisions.

Best for:

  • Legal and scientific analysis
  • Complex multi-step reasoning
  • Agentic coding workflows
  • High-stakes decision support
Full pricing breakdown →
Budget

Claude Haiku 3.5

$0.80 / $4 input / output per MTok

Fast and affordable. Purpose-built for high-volume workloads where speed and cost matter more than deep reasoning.

Best for:

  • Classification and routing
  • Entity extraction
  • Content moderation
  • Real-time data formatting
Full pricing breakdown →

How Does Claude Compare?

See how Claude API pricing stacks up against the other major LLM providers, token-for-token.

Frequently Asked Questions

How much does the Claude API cost?
Claude API pricing depends on the model you choose. Claude Haiku 3.5 costs $0.80 per million input tokens and $4.00 per million output tokens. Claude Sonnet 4 costs $3.00/$15.00, and Claude Opus 4 costs $15.00/$75.00. You only pay for the tokens you use, with no minimum commitment or monthly fee.
What is the cheapest Claude model?
Claude Haiku 3.5 is the cheapest model at $0.80 per million input tokens and $4.00 per million output tokens. For even lower costs, combine Haiku with the Batch API (50% off) to get input pricing as low as $0.40 per million tokens.
How do I estimate my Claude API costs?
Use our cost calculator. Enter your model, average input and output tokens per request, and requests per day. A rough rule of thumb: 1 token is about 4 characters of English text, and 1 million tokens is approximately 750,000 words or 1,500 pages of standard text.
What is prompt caching and how much does it save?
Prompt caching lets you cache repeated parts of your input (system prompts, shared context, few-shot examples) and pay only 10% of the normal input price for cached tokens. The initial cache write costs 25% more than the base input price. If 70% of your input is cacheable, you save roughly 63% on input costs after the first request.
What is the Claude Batch API?
The Batch API gives you a flat 50% discount on both input and output tokens for workloads that do not need real-time responses. You submit a JSONL file of requests, and results are processed within a 24-hour window. It is ideal for content generation, bulk classification, data processing, and evaluation pipelines.
Is the Claude API cheaper than OpenAI?
It depends on the model tier. Claude Sonnet 4 ($3/$15) costs more per token than GPT-4o ($2.50/$10), but prompt caching can reduce Claude input costs by up to 90%, making it more affordable for workloads with repeated context. Claude Haiku 3.5 ($0.80/$4) is pricier than GPT-4o-mini ($0.15/$0.60) per token but delivers stronger performance on complex tasks.
How many tokens are in a typical API request?
One token is roughly 4 characters or 0.75 words of English text. A typical chatbot message uses 500-2,000 input tokens and 200-1,000 output tokens. A document summarisation request with a full document might use 4,000-30,000 input tokens and 500-1,500 output tokens. Code generation usually runs 1,000-4,000 input and 1,000-3,000 output tokens.
Does Claude have a free tier?
Anthropic provides limited free API credits for evaluation when you create a new account. Beyond that, you pay per token with no minimum spend. There is no permanent free tier for production use. Enterprise customers can negotiate volume discounts and custom commitments with Anthropic's sales team.