Struct Completion 

#[non_exhaustive]
pub struct Completion {
    pub text: String,
    pub json: Option<Value>,
    pub model: String,
    pub input_tokens: u64,
    pub output_tokens: u64,
    pub cached_input_tokens: Option<u64>,
}

A completed completion: the text, the model that answered (a provider identifier, for the log line), and the token usage.

The two token counts have a fixed relationship: Completion::input_tokens is the total prompt tokens the provider processed, and Completion::cached_input_tokens — where the provider reports one — is the subset of that total served from the provider’s own prompt cache. A caller costing a completion never subtracts; the total already includes the cached part.

#[non_exhaustive] for the same reason Prompt is: what a provider reports back grows, and it should not break every caller.

Fields (Non-exhaustive)

Non-exhaustive structs could have additional fields added in future. Therefore, non-exhaustive structs cannot be constructed in external crates using the traditional Struct { .. } syntax; cannot be matched against without a wildcard ..; and struct update syntax will not work.

text: String

json: Option<Value>

The parsed answer, when the prompt carried a Prompt::json_schema. None when it did not, or the provider’s answer could not be parsed — in which case text still holds what came back.

model: String

input_tokens: u64

The total prompt tokens the provider processed, cached tokens included — a provider that reports caching reports them alongside the uncached remainder, not inside it. The discount question is Completion::cached_input_tokens, never a smaller total here.

output_tokens: u64

cached_input_tokens: Option<u64>

The subset of Completion::input_tokens the provider served from its own prompt cache, where the provider reports one. None when it does not: an absent report and a reported zero mean different things (a vendor that does not do caching versus a cache miss), so the option is never collapsed.

Implementations

impl Completion

pub fn new(text: impl Into<String>, model: impl Into<String>) -> Self

A completion with just the text and the model that wrote it; the usage and parsed JSON are builder methods.

pub fn json(self, json: Value) -> Self

The parsed structured answer, for a prompt that asked for one.

pub fn usage(self, input_tokens: u64, output_tokens: u64) -> Self

The token counts the provider reported.

pub fn cached_input_tokens(self, cached_input_tokens: u64) -> Self

The subset of the input tokens the provider served from its own prompt cache, where the provider reports one. Completion::input_tokens stays the total, cached included.

Trait Implementations

impl Clone for Completion

fn clone(&self) -> Self

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for Completion

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

Formats the value using the given formatter.

impl PartialEq for Completion

fn eq(&self, other: &Self) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for Completion