Struct ContextSizing 

pub struct ContextSizing {
    pub model_max: Option<usize>,
    pub effective: Option<usize>,
    pub source: Option<NumCtxSource>,
    pub max_output: Option<usize>,
    pub compacts_natively: bool,
}

Resolved context sizing for a turn. For most providers model_max == effective (the static advertised window). For Ollama they differ: model_max is the probed architectural window, while effective is what we actually enforce as num_ctx (auto-fitted to memory, capped, or an override). Compaction and the status bar use effective; “model supports up to X” uses model_max.

Fields

model_max: Option<usize>

effective: Option<usize>

source: Option<NumCtxSource>

How effective was chosen (Ollama only). None for static/advertised.

max_output: Option<usize>

The model’s per-response output ceiling, when the provider exposes one (/models metadata, or a documented static table). Rides the same resolve→reducer pipeline as the window so provider_capabilities can be refreshed live.

compacts_natively: bool

Whether the provider compacts the conversation itself when a turn asks (ChatRequest::native_compaction): Anthropic, until the model refuses.

Trait Implementations

impl Clone for ContextSizing

fn clone(&self) -> Self

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Copy for ContextSizing

impl Debug for ContextSizing

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

Formats the value using the given formatter.

impl Default for ContextSizing

fn default() -> Self

Returns the “default value” for a type.