Struct ModelResponse 

pub struct ModelResponse {
    pub id: String,
    pub model: String,
    pub stop_reason: StopReason,
    pub content: Vec<ContentPart>,
    pub usage: Usage,
    pub rate_limit: Option<RateLimitSnapshot>,
    pub warnings: Vec<ModelWarning>,
    pub provider_echoes: Vec<ProviderEchoSnapshot>,
}

One reply from a model invocation, after decoding.

Fields

id: String

Vendor-assigned response ID (used for tracing and replay).

model: String

Echo of the model that produced this response — useful when the codec resolved an alias (e.g. claude-opus → claude-opus-4-7-20260415).

stop_reason: StopReason

Why the model stopped producing tokens.

content: Vec<ContentPart>

Returned content blocks (text, tool calls, etc.).

usage: Usage

Token / cache accounting from the vendor.

rate_limit: Option<RateLimitSnapshot>

Provider rate-limit state at response time, when the codec could extract it from response headers (Codec::extract_rate_limit).

warnings: Vec<ModelWarning>

Codec-emitted warnings (lossy encoding, unknown stop reasons, etc.). Always non-fatal; consumers may surface them in observability.

provider_echoes: Vec<ProviderEchoSnapshot>

Vendor-keyed opaque round-trip tokens that ride at the response root (rather than on a single content part) — OpenAI Responses Response.id is the canonical example, captured here so the next ModelRequest::continued_from can chain via previous_response_id. Codecs only populate entries matching their own Codec::name.

Implementations

impl ModelResponse

pub fn first_text(&self) -> Option<&str>

Borrow the first text block, if any. Convenient when the model is expected to reply with a single text answer (the 5-line-agent path) — saves the manual match &response.content[0] { ContentPart::Text { text, .. } => …, _ => panic!() } dance at every call site.

Returns None when the response has no text block (e.g. the reply is purely a ToolUse block).

pub fn full_text(&self) -> String

Concatenate every text block in order. Useful when the model emits multiple Text blocks interleaved with Thinking or ToolUse blocks and the caller only wants the user-visible answer string.

pub fn tool_uses(&self) -> Vec<ToolUseRef<'_>>

Borrow every ToolUse block in declaration order. Empty when the response carried no tool calls. Used by ReAct-style agents to drive the next dispatch round.

pub fn has_tool_uses(&self) -> bool

True iff the response has at least one ToolUse block — hot path for agent loops that branch on “did the model ask to call a tool?”.

Trait Implementations

impl Clone for ModelResponse

fn clone(&self) -> ModelResponse

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for ModelResponse

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for ModelResponse

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl PartialEq for ModelResponse

fn eq(&self, other: &ModelResponse) -> 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 ModelResponse

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

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl Eq for ModelResponse

impl StructuralPartialEq for ModelResponse