Blog Studio Go to console

FirebaseAILogic Framework Reference

UsageMetadata

@available(iOS 15.0, macOS 12.0, tvOS 15.0, watchOS 8.0, *)
public struct UsageMetadata : Sendable
 
extension GenerateContentResponse.UsageMetadata: Decodable

Token usage metadata for processing the generate content request.

  • The number of tokens in the request prompt.

    Declaration

    Swift

    public let promptTokenCount: Int

  • The total number of tokens across the generated response candidates.

    Declaration

    Swift

    public let candidatesTokenCount: Int

  • The number of tokens used by tools.

    Declaration

    Swift

    public let toolUsePromptTokenCount: Int

  • The number of tokens used by the model’s internal “thinking” process.

    For models that support thinking (like Gemini 2.5 Pro and Flash), this represents the actual number of tokens consumed for reasoning before the model generated a response. For models that do not support thinking, this value will be 0.

    When thinking is used, this count will be less than or equal to the thinkingBudget set in the ThinkingConfig.

    Declaration

    Swift

    public let thoughtsTokenCount: Int

  • The total number of tokens in both the request and response.

    Declaration

    Swift

    public let totalTokenCount: Int

  • The breakdown, by modality, of how many tokens are consumed by the prompt.

    Declaration

    Swift

    public let promptTokensDetails: [ModalityTokenCount]

  • The breakdown, by modality, of how many tokens are consumed by the candidates

    Declaration

    Swift

    public let candidatesTokensDetails: [ModalityTokenCount]

  • The breakdown, by modality, of how many tokens were consumed by the tools used to process the request.

    Declaration

    Swift

    public let toolUsePromptTokensDetails: [ModalityTokenCount]

Codable Conformances

  • Declaration

    Swift

    public init(from decoder: any Decoder) throws