Struct GainSummary 

pub struct GainSummary {

    pub model: ModelQuote,
    pub total_commands: u64,
    pub input_tokens: u64,
    pub output_tokens: u64,
    pub tokens_saved: u64,
    pub effective_input_tokens: u64,
    pub effective_output_tokens: u64,
    pub effective_tokens_saved: u64,
    pub effective_compression_pct: f64,
    pub total_reduction_pct: f64,
    pub gain_rate_pct: f64,
    pub injected_overhead_tokens_per_turn: u64,
    pub turns: u64,
    pub injected_overhead_total_tokens: u64,
    pub net_tokens_saved: i64,
    pub injected_overhead_budget_tokens: u64,
    pub over_budget: bool,
    pub avoided_usd: f64,
    pub stream_savings: StreamSavings,
    pub energy_wh: f64,
    pub co2_grams: f64,
    pub tool_spend_usd: f64,
    pub roi: Option<f64>,
    pub score: GainScore,
    pub daemon_hint: Option<String>,
}

Fields

model: ModelQuote

total_commands: u64

input_tokens: u64

output_tokens: u64

tokens_saved: u64

effective_input_tokens: u64

Tokens removed from traffic whose output lean-ctx controls.

effective_output_tokens: u64

effective_tokens_saved: u64

effective_compression_pct: f64

Compression percentage over compressible traffic only.

total_reduction_pct: f64

Reduction percentage over all recorded traffic, including passthrough.

gain_rate_pct: f64

injected_overhead_tokens_per_turn: u64

Fixed per-turn context lean-ctx injects (tool schemas + server instructions + rules block). On a provider WITHOUT prompt caching this rides — and is re-billed on — every turn, so the net bill impact is tokens_saved − injected_overhead_tokens_per_turn × turns. Note that tokens_saved / gain_rate_pct are measured against lean-ctx-touched traffic (their denominator), not the full provider bill (GitHub #361).

turns: u64

Provider turns (requests) the proxy actually saw carry the injected prefix. 0 when the proxy is not in the request path, in which case the net figure below collapses to the gross tokens_saved (we cannot count turns we never observed, and we refuse to guess).

injected_overhead_total_tokens: u64

injected_overhead_tokens_per_turn × turns — the total fixed context tax re-billed across the run on a provider without prompt caching.

net_tokens_saved: i64

The honest bill impact: tokens_saved − injected_overhead_total_tokens. Signed, because on a non-caching rail a short run can legitimately go net-negative until savings outgrow the per-turn injection.

injected_overhead_budget_tokens: u64

Configured fixed-context budget ([context] budget_tokens); 0 disables the check (#964). Surfaced so gain can flag a bloated injected prefix.

over_budget: bool

Whether injected_overhead_tokens_per_turn exceeds a non-zero budget (#964).

avoided_usd: f64

stream_savings: StreamSavings

Bill-weighted savings split by provider token stream (#1191).

energy_wh: f64

Estimated grid energy avoided (Wh) by keeping tokens_saved out of context.

co2_grams: f64

Estimated CO₂-equivalent avoided (grams), derived from energy_wh.

tool_spend_usd: f64

roi: Option<f64>

score: GainScore

daemon_hint: Option<String>

Trait Implementations

impl Clone for GainSummary

fn clone(&self) -> GainSummary

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for GainSummary

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for GainSummary

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for GainSummary

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

where __S: Serializer,

Serialize this value into the given Serde serializer.