Struct ModelRequest 

pub struct ModelRequest {
    pub model: String,
    pub max_output_tokens: Option<u32>,
    pub thinking_level: ThinkingLevel,
    /* private fields */
}

A complete, attributed model request. Constructed only via ModelRequestBuilder.

Fields

model: String

max_output_tokens: Option<u32>

thinking_level: ThinkingLevel

Implementations

impl ModelRequest

pub fn builder(model: impl Into<String>) -> ModelRequestBuilder

pub fn fragments(&self) -> &[PromptFragment]

pub fn tools(&self) -> &[ToolDefinition]

pub fn bytes_by_source(&self) -> BTreeMap<ContentSource, usize>

Byte attribution by source category, including tool definitions. The heart of the attribution invariant: this is derivable for every request because no content can bypass fragment construction.

pub fn estimate(&self) -> RequestEstimate

Categorized heuristic estimate. Tokenizer counts and provider-reported usage are separate measurements; this does not count wire framing.

pub fn total_bytes(&self) -> usize

pub fn with_suffix_instruction( &self, source: ContentSource, instruction: &str, ) -> Self

Prefix-cache pattern: the live request, unchanged, with one instruction appended as a suffix (compaction summaries, titles). Tool definitions are dropped — suffix requests want plain text back.

pub fn with_warm_probe(&self) -> Self

Cache-warm probe: the request unchanged — tool definitions kept, they are part of the provider’s cached prefix (Anthropic caches tools → system → messages) — plus one throwaway user fragment (providers reject empty message lists). The fragment sits after the system-block cache breakpoint, so its content never affects the cached prefix. max_output_tokens: 1 caps waste if the caller’s first-event abort is slow.

pub fn without_image_blocks(&self) -> Self

The request with every image block replaced by a text note, for models whose endpoint accepts text content only — sending the image block would fail the whole request at the provider. The session log is untouched: history keeps the real image parts, so switching to an image-capable model replays them as images again. Attribution is preserved per fragment; only the image bytes become the note’s text.

Trait Implementations

impl Clone for ModelRequest

fn clone(&self) -> ModelRequest

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for ModelRequest

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

Formats the value using the given formatter.

impl JsonSchema for ModelRequest

fn schema_name() -> Cow<'static, str>

The name of the generated JSON Schema.

fn schema_id() -> Cow<'static, str>

Returns a string that uniquely identifies the schema produced by this type.

fn json_schema(generator: &mut SchemaGenerator) -> Schema

Generates a JSON Schema for this type.

fn inline_schema() -> bool

Whether JSON Schemas generated for this type should be included directly in parent schemas, rather than being re-used where possible using the $ref keyword.

impl PartialEq for ModelRequest

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

Tests for self and other values to be equal, and is used by ==.

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

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Serialize for ModelRequest

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

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl StructuralPartialEq for ModelRequest