Struct ChatUsage 

pub struct ChatUsage {
    pub input_tokens: i32,
    pub output_tokens: i32,
    pub cost_ticks: i64,
    pub cached_tokens: Option<i64>,
    pub reasoning_tokens: Option<i64>,
}

Token counts and cost for a chat response.

Fields

input_tokens: i32

output_tokens: i32

cost_ticks: i64

cached_tokens: Option<i64>

Input tokens served from the provider’s prompt cache, billed at the (lower) cached rate. Omitted on responses with no cache hit. Backend wire shape: cached_tokens (i32); promoted to Option here for headroom on future long-cache scenarios (multi-hour video transcripts, etc.). Multi-turn billing audits reconcile by computing (non-cached) input_tokens vs (cached) cached_tokens — both roll into the provider’s billable total.

reasoning_tokens: Option<i64>

Sub-component of output_tokens that was model reasoning / thinking output (Gemini 3.x’s thoughtTokens, OpenAI o- series’ reasoning tokens, Anthropic’s extended-thinking blocks). Omitted on responses from non-reasoning models. Total billable output = output_tokens (which already includes the reasoning component — this field is just transparency on how much of that was thinking).

Trait Implementations

impl Clone for ChatUsage

fn clone(&self) -> ChatUsage

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for ChatUsage

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for ChatUsage

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.