Struct TokenUsage 

pub struct TokenUsage {
    pub prompt_tokens: u64,
    pub completion_tokens: u64,
    pub cache_read_tokens: u64,
    pub cache_creation_tokens: u64,
    pub thinking_tokens: u64,
    pub count_type: TokenCountType,
    pub request_count: u64,
    pub session_start: Option<Instant>,
}

Token usage statistics for /cost command Tracks actual vs approximate tokens similar to Forge

Fields

prompt_tokens: u64

Total prompt/input tokens

completion_tokens: u64

Total completion/output tokens

cache_read_tokens: u64

Cache read tokens (prompt caching)

cache_creation_tokens: u64

Cache creation tokens (prompt caching)

thinking_tokens: u64

Thinking/reasoning tokens (extended thinking models)

count_type: TokenCountType

Whether the counts are actual or approximate

request_count: u64

Number of requests made

session_start: Option<Instant>

Session start time

Implementations

impl TokenUsage

pub fn new() -> Self

pub fn add_actual(&mut self, input: u64, output: u64)

Add actual tokens from API response

pub fn add_actual_extended( &mut self, input: u64, output: u64, cache_read: u64, cache_creation: u64, thinking: u64, )

Add actual tokens with cache and thinking info

pub fn add_estimated(&mut self, prompt: u64, completion: u64)

Add estimated tokens (when API doesn’t return actual counts) Only updates if we don’t already have actual counts for this session

pub fn add_request(&mut self, prompt: u64, completion: u64)

Legacy method for compatibility - adds estimated tokens

pub fn estimate_tokens(text: &str) -> u64

Estimate token count from text (rough approximation: ~4 chars per token) Matches Forge’s approach: char_count.div_ceil(4)

pub fn total_tokens(&self) -> u64

Get total tokens (input + output, excluding cache/thinking)

pub fn total_with_cache(&self) -> u64

Get total tokens including cache reads (effective context size)

pub fn format_total(&self) -> String

Format total tokens for display (with ~ prefix if approximate)

pub fn format_compact(&self) -> String

Get a short display string like Forge: “~1.2k” or “15k”

pub fn has_cache_hits(&self) -> bool

Check if we have cache hits (prompt caching is working)

pub fn has_thinking(&self) -> bool

Check if we have thinking tokens (extended thinking enabled)

pub fn session_duration(&self) -> Duration

Get session duration

pub fn estimate_cost(&self, model: &str) -> (f64, f64, f64)

Estimate cost based on model (rough estimates in USD) Returns (input_cost, output_cost, total_cost)

pub fn print_report(&self, model: &str)

Print cost report

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

Formats the value using the given formatter.

impl Default for TokenUsage

fn default() -> TokenUsage

Returns the “default value” for a type.