google-adk-kotlin-core/com.google.adk.kt.types/UsageMetadata

UsageMetadata

@Serializable
data class UsageMetadata(val promptTokenCount: Int? = null, val candidatesTokenCount: Int? = null, val totalTokenCount: Int? = null, val thoughtsTokenCount: Int? = null, val toolUsePromptTokenCount: Int? = null, val cachedContentTokenCount: Int? = null, val promptTokensDetails: List<ModalityTokenCount>? = null, val candidatesTokensDetails: List<ModalityTokenCount>? = null)

Usage metadata for a generate content request.

Constructors

Link copied to clipboard
constructor(promptTokenCount: Int? = null, candidatesTokenCount: Int? = null, totalTokenCount: Int? = null, thoughtsTokenCount: Int? = null, toolUsePromptTokenCount: Int? = null, cachedContentTokenCount: Int? = null, promptTokensDetails: List<ModalityTokenCount>? = null, candidatesTokensDetails: List<ModalityTokenCount>? = null)

Properties

Link copied to clipboard
val cachedContentTokenCount: Int? = null

The number of tokens served from the cached content (cache read).

Link copied to clipboard
val candidatesTokenCount: Int? = null

The number of tokens in the candidates.

Link copied to clipboard
val candidatesTokensDetails: List<ModalityTokenCount>? = null

A per-modality breakdown of the candidates token count.

Link copied to clipboard
val promptTokenCount: Int? = null

The number of tokens in the prompt.

Link copied to clipboard
val promptTokensDetails: List<ModalityTokenCount>? = null

A per-modality breakdown of the prompt token count.

Link copied to clipboard
val thoughtsTokenCount: Int? = null

The number of tokens that were part of the model's "thoughts" output, for thinking models.

Link copied to clipboard
val toolUsePromptTokenCount: Int? = null

The number of tokens in tool-execution results provided back to the model as input.

Link copied to clipboard
val totalTokenCount: Int? = null

The total number of tokens.