Token Usage

Tokens are the base unit models use to process text. The actual count depends on the model tokenizer and the full request context.

Reading Usage from Responses

Non-streaming responses return input, output, and total tokens in the usage field. Whether streaming requests return usage depends on the upstream model and request parameters.

{
  "usage": {
    "prompt_tokens": 128,
    "completion_tokens": 42,
    "total_tokens": 170
  }
}