Struct CompletionResponse 

pub struct CompletionResponse {

    pub id: String,
    pub object: ResponseObject,
    pub created_at: u64,
    pub status: ResponseStatus,
    pub error: Option<ResponseError>,
    pub incomplete_details: Option<IncompleteDetailsReason>,
    pub instructions: Option<String>,
    pub max_output_tokens: Option<u64>,
    pub model: String,
    pub provider_reasoning: Option<String>,
    pub provider_request_id: Option<String>,
    pub reasoning_metadata: Option<Map<String, Value>>,
    pub reasoning_context: Option<String>,
    pub usage: Option<ResponsesUsage>,
    pub output: Vec<Output>,
    pub tools: Vec<ResponsesToolDefinition>,
    pub additional_parameters: AdditionalParameters,
}

The standard response format from OpenAI’s Responses API.

Fields

id: String

The ID of a completion response.

object: ResponseObject

The type of the object.

created_at: u64

The time at which a given response has been created, in seconds from the UNIX epoch (01/01/1970 00:00:00).

status: ResponseStatus

The status of the response.

error: Option<ResponseError>

Response error (optional)

incomplete_details: Option<IncompleteDetailsReason>

Incomplete response details (optional)

instructions: Option<String>

System prompt/preamble

max_output_tokens: Option<u64>

The maximum number of tokens the model should output

model: String

The model name

provider_reasoning: Option<String>

Provider-specific top-level reasoning content returned by some OpenAI-compatible Responses implementations.

provider_request_id: Option<String>

The transport request id from the x-request-id response header — not part of the response body; stamped by the request driver, so wire deserialization always leaves it None and the manual Serialize (which mirrors the wire body) never emits it.

reasoning_metadata: Option<Map<String, Value>>

The complete object-shaped top-level reasoning metadata returned by the provider.

Unknown fields, unknown values, and null-valued members inside the object are preserved value-equivalently. A top-level null, missing field, or unsupported non-object shape is normalized to no reasoning metadata. When serializing manually constructed responses, Self::provider_reasoning takes precedence over this field, and this field takes precedence over Self::reasoning_context.

reasoning_context: Option<String>

The effective reasoning context returned by OpenAI.

This is populated as a convenience projection of Self::reasoning_metadata. String-shaped reasoning returned by compatible providers remains available through Self::provider_reasoning.

usage: Option<ResponsesUsage>

Token usage

output: Vec<Output>

The model output (messages, etc will go here)

tools: Vec<ResponsesToolDefinition>

Tools

additional_parameters: AdditionalParameters

Additional parameters

Trait Implementations

impl Clone for CompletionResponse

fn clone(&self) -> CompletionResponse

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for CompletionResponse

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for CompletionResponse

fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>

where D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl NormalizeCompletionResponse for CompletionResponse

Normalize an OpenAI Responses API completion.

The provider descriptor name is an input rather than a constant: ChatGPT and Copilot return this exact wire shape, so hardcoding "openai" here would mislabel them. Taking it as part of the conversion makes the correct name impossible to forget.

fn normalize( self, provider: &str, ) -> Result<CompletionResponse, CompletionError>

Normalize this payload, attributing it to provider.

impl ProviderResponseExt for CompletionResponse

fn get_response_id(&self) -> Option<String>

The response ID (resp_...), which is deliberately not the assistant message ID (msg_...) that the normalized response carries.

type Usage = ResponsesUsage

Provider-native usage type.

fn get_response_model_name(&self) -> Option<String>

Returns the provider response model name, if supplied.

fn get_text_response(&self) -> Option<String>

Returns the primary text response, when available.

fn get_usage(&self) -> Option<Self::Usage>

Returns provider-native usage metrics, if supplied.

impl Serialize for CompletionResponse

fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>

where S: Serializer,

Serialize this value into the given Serde serializer.