Struct TokenUsage 

pub struct TokenUsage {
    pub model: String,
    pub input_tokens: u64,
    pub output_tokens: u64,
    pub cached_input_tokens: u64,
    pub cache_creation_input_tokens: u64,
    pub total_tokens: u64,
    pub cost_usd: f64,
    pub unpriced_tokens: u64,
    pub pricing_available: bool,
    pub timestamp: DateTime<Utc>,
}

Fields

model: String

Model identifier (e.g., “anthropic/claude-sonnet-4-20250514”)

input_tokens: u64

Input/prompt tokens

output_tokens: u64

Output/completion tokens

cached_input_tokens: u64

Cached input tokens (Anthropic cache_read_input_tokens, OpenAI prompt_tokens_details.cached_tokens). Subset of input_tokens when reported by the provider; the rate sheet’s cached_input_per_mtok applies to these.

cache_creation_input_tokens: u64

Cache-write tokens (Anthropic cache_creation_input_tokens, OpenAI-compatible prompt_tokens_details.cache_creation_input_tokens). Subset of the uncached input that the provider wrote into its prompt cache on this request; the rate sheet’s cache_write_per_mtok applies to these when set, otherwise they stay priced at the plain input rate.

total_tokens: u64

Total tokens (input + output, ignoring the cached subset).

cost_usd: f64

Calculated cost in USD

unpriced_tokens: u64

Token-bearing dimensions that had no valid resolved price when this record was created. This is the canonical exposure count for new rows; pricing_available remains as a compatibility projection for older ledgers written before dimension-level provenance existed.

pricing_available: bool

timestamp: DateTime<Utc>

Timestamp of the request

Implementations

impl TokenUsage

pub fn billable_input_tokens(&self) -> u64

pub fn new( model: impl Into<String>, input_tokens: u64, output_tokens: u64, cached_input_tokens: u64, input_price_per_million: f64, output_price_per_million: f64, cached_input_price_per_million: f64, ) -> TokenUsage

pub fn new_with_cache( model: impl Into<String>, input_tokens: u64, cached_input_tokens: u64, output_tokens: u64, input_price_per_million: f64, cached_input_price_per_million: f64, output_price_per_million: f64, ) -> TokenUsage

Create a new token usage record with cache-aware input pricing.

pub fn new_with_cache_write( model: impl Into<String>, input_tokens: u64, cached_input_tokens: u64, cache_creation_input_tokens: u64, output_tokens: u64, input_price_per_million: f64, cached_input_price_per_million: f64, cache_write_price_per_million: f64, output_price_per_million: f64, ) -> TokenUsage

Create a new token usage record that also prices the cache-write band: cache_creation_input_tokens at cache_write_price_per_million. A non-positive write rate keeps the historical behavior — cache writes stay inside the plain uncached-input band at the input rate — so sheets without a write rate reproduce yesterday’s totals exactly.

pub fn cost(&self) -> f64

Get the total cost.

Trait Implementations

impl Clone for TokenUsage

fn clone(&self) -> TokenUsage

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for TokenUsage

fn fmt(&self, f: &mut Formatter<'_>) -> Result<(), Error>

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for TokenUsage

fn deserialize<__D>( __deserializer: __D, ) -> Result<TokenUsage, <__D as Deserializer<'de>>::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for TokenUsage

fn serialize<__S>( &self, __serializer: __S, ) -> Result<<__S as Serializer>::Ok, <__S as Serializer>::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.