Claude API pricing
Claude Token Cost Calculator
Estimate monthly Anthropic API spend from Claude model choice, request volume, input tokens, output tokens, prompt caching, and deployment geography.
Estimate Claude token cost
Plan by request volume
Turn average input and output tokens into monthly Claude API cost and cost per request.
Model cache hits and writes
Separate standard input, cache reads, and cache writes so repeated prompts are not overestimated.
Add geography and buffer
Account for US-only inference pricing and add a buffer for retries, prompt growth, and traffic spikes.
How Claude token cost is calculated
Monthly cost is estimated from standard input tokens, cache read tokens, cache write tokens, output tokens, the selected Claude model's prices, and optional geography multiplier.
Formula: standard input + cache reads + cache writes + output. The budget estimate adds your selected buffer percentage.
| Planning question | Use this input | Related page |
|---|---|---|
| How many Claude tokens does my prompt use? | Estimate prompt size before setting the input token field. | Claude Tokenizer |
| How much does prompt caching change cost? | Adjust cacheable input percent and cache hit percent. | Prompt Caching Savings Calculator |
| What if I compare Claude with OpenAI or Gemini? | Use the cross-provider LLM calculator. | LLM Cost Calculator |
| What if I only need Gemini cost? | Use Gemini-specific caching and grounding assumptions. | Gemini Cost Calculator |
FAQ
Why does prompt caching have read and write costs?
Claude prompt caching charges a cache write when content is stored and a cache read when a later request retrieves that cached content.
Should I use input tokens from this page or the Claude tokenizer page?
Use this page when you already have average input and output token counts. Use the Claude tokenizer page first when you only have raw prompt text.
Sources
Claude prices in this site were last checked on 2026-07-02. Pricing, model availability, and regional multipliers can change.