Struct CallOptions 

pub struct CallOptions {

    pub prompt: Prompt,
    pub max_output_tokens: Option<u32>,
    pub temperature: Option<f32>,
    pub stop_sequences: Option<Vec<String>>,
    pub top_p: Option<f32>,
    pub top_k: Option<u32>,
    pub presence_penalty: Option<f32>,
    pub frequency_penalty: Option<f32>,
    pub response_format: Option<ResponseFormat>,
    pub seed: Option<i64>,
    pub tools: Option<Vec<Tool>>,
    pub tool_choice: Option<ToolChoice>,
    pub include_raw_chunks: Option<bool>,
    pub headers: Option<HashMap<String, String>>,
    pub provider_options: Option<HashMap<String, HashMap<String, JsonValue>>>,
}

Configuration options for language model generation requests.

This struct encapsulates all available parameters for controlling language model behavior. It provides fine-grained control over sampling, generation limits, response formatting, and tool integration. Use this structure when making requests to language models through the SDK.

Fields

• prompt - The input prompt or conversation history to send to the model. This is the primary content that guides the model’s response generation.

• max_output_tokens - The maximum number of tokens the model should generate in its response. Helps control response length and API costs. Typical values range from 100 to 4096.

• temperature - Controls randomness in token selection. Values from 0.0 to 2.0:

  • 0.0: Deterministic, always selects the most likely token
  • 0.7: Balanced randomness (recommended for most use cases)
  • 2.0: Maximum randomness

• stop_sequences - List of strings that, when generated, will cause the model to stop generating further tokens. Useful for controlling output boundaries.

• top_p - Nucleus sampling parameter (0.0 to 1.0). The model selects from the smallest set of tokens whose probabilities sum to this value. Works well with temperature.

• top_k - Top-k sampling parameter. Limits selection to the k most likely next tokens, reducing tail probability mass. A value of 40-50 is typical.

• presence_penalty - Penalty for tokens that have already appeared in the prompt (-2.0 to 2.0). Positive values encourage diversity and reduce repetition.

• frequency_penalty - Penalty based on token frequency in the generated text (-2.0 to 2.0). Positive values reduce likelihood of repeating common phrases.

• response_format - Specifies the desired format for the model’s response, either plain text or structured JSON with optional schema validation.

• seed - Random seed for deterministic output. When set, the same seed with the same inputs will produce identical outputs (if supported by the provider).

• tools - Array of tool definitions available for the model to call. Enables function calling and tool use capabilities.

• tool_choice - Strategy for tool invocation: Auto (model decides), None (disable tools), Required (must call at least one), or a specific tool by name.

• include_raw_chunks - When streaming, includes raw provider chunks in the response for debugging or advanced use cases.

• headers - Additional HTTP headers to include in the request. Useful for custom authentication, routing, or provider-specific headers.

• provider_options - Provider-specific configuration options that may not be part of the standard SDK parameters.

Fields

prompt: Prompt

The input prompt or conversation history for the model.

max_output_tokens: Option<u32>

Maximum number of tokens to generate in the response.

temperature: Option<f32>

Sampling temperature controlling randomness (0.0 to 2.0).

stop_sequences: Option<Vec<String>>

Sequences that trigger generation to stop when encountered.

top_p: Option<f32>

Nucleus sampling probability mass (0.0 to 1.0).

top_k: Option<u32>

Top-k sampling parameter limiting token selection.

presence_penalty: Option<f32>

Penalty for tokens that have appeared in the prompt (-2.0 to 2.0).

frequency_penalty: Option<f32>

Penalty based on token frequency in generated text (-2.0 to 2.0).

response_format: Option<ResponseFormat>

Expected format for the model’s response (Text or JSON).

seed: Option<i64>

Random seed for deterministic generation.

tools: Option<Vec<Tool>>

Tool definitions available for the model to call.

tool_choice: Option<ToolChoice>

Strategy for tool selection and invocation.

include_raw_chunks: Option<bool>

Whether to include raw provider chunks in streaming responses.

headers: Option<HashMap<String, String>>

Additional HTTP headers for the provider request.

provider_options: Option<HashMap<String, HashMap<String, JsonValue>>>

Provider-specific configuration options.

Trait Implementations

impl Clone for CallOptions

fn clone(&self) -> CallOptions

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for CallOptions

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

Formats the value using the given formatter.

impl Default for CallOptions

fn default() -> CallOptions

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for CallOptions

fn deserialize<__D>( __deserializer: __D, ) -> Result<CallOptions, <__D as Deserializer<'de>>::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl PartialEq for CallOptions

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

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

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl StructuralPartialEq for CallOptions