Token Usage

Tokens are the fundamental units used by language models to represent text. They are also the units used for metering and billing. A token typically corresponds to a short word, a symbol, a number, or a character—depending on the language and encoding.

For intuitive reference:

• A Chinese character usually counts as ~0.6 tokens

• An English character typically equals ~0.3 tokens

• A single English word or number often maps to 1 token

⚠️ Note: The exact token count depends on the model’s tokenizer. Different models may tokenize the same input differently. Final token usage is always based on what the model returns per request.