Billing Overview
What Is a Token?
A token is the basic unit models use to process text. As a rough estimate:
- 1 English word is about 1 to 2 tokens
- 1 Chinese character is about 1.5 to 2 tokens
How Is Billing Calculated?
Claude model token usage is divided into four categories:
| Token type | Description |
|---|---|
| Input tokens | The content you send to the model, such as prompts, code, and context |
| Output tokens | The response content generated by the model |
| Cache write tokens | Input content cached for the first time. This is more expensive than standard input, but later reuse can save significantly. |
| Cache read tokens | Input content hit from cache. This costs much less than standard input. |
Complete billing formula:
Cost = input tokens × input price
+ output tokens × output price
+ cache write tokens × cache write price
+ cache read tokens × cache read price
About caching In tools such as Claude Code, the system automatically manages context caching. When you keep chatting in the same session, a large amount of input content is cached, and cache read tokens in later requests cost far less than regular input. This can significantly reduce costs. That is also why we recommend using the corresponding group in each tool: it makes cache hits easier.