Struct LanguageModelRequest

Source

pub struct LanguageModelRequest<M>where
    M: LanguageModel,{
    pub model: M,
    pub prompt: Option<String>,
    /* private fields */
}

Expand description

Options for text generation requests such as generate_text and stream_text.

Fields§

§model: M

The language model to use for text generation.

§prompt: Option<String>

An optional simple text prompt for the request.

This should not be set if messages are provided in the options.

Implementations§

Source §

impl<M> LanguageModelRequest<M>
where M: LanguageModel,

Source

pub async fn generate_text(&mut self) -> Result<GenerateTextResponse, Error>

Generates text and executes tools using the language model.

This method performs non-streaming text generation, potentially involving multiple steps of tool calling and execution until the conversation reaches a natural stopping point. The model may call tools based on the configured options, and responses are processed iteratively until completion.

For streaming responses, use stream_text instead.

§Returns

A GenerateTextResponse containing the final conversation state and generated content.

§Errors

Returns an Error if the underlying language model fails to generate a response or if tool execution encounters an error.

§Examples

   use aisdk::{
       core::{LanguageModelRequest},
       providers::OpenAI,
   };

   async fn main() -> Result<(), Box<dyn std::error::Error>> {

       let openai = OpenAI::gpt_5();

       let result = LanguageModelRequest::builder()
           .model(openai)
           .prompt("What is the meaning of life?")
           .build()
           .generate_text()
           .await?;

       println!("{}", result.text().unwrap());
       Ok(())
   }

Source §

impl<M> LanguageModelRequest<M>
where M: LanguageModel,

Source

pub fn builder() -> LanguageModelRequestBuilder<M>

Creates a new builder for constructing a LanguageModelRequest.

This method initiates the type-state builder pattern, starting with the ModelStage where you must specify the language model.

Source §

impl<M> LanguageModelRequest<M>
where M: LanguageModel,

Source

pub async fn stream_text(&mut self) -> Result<StreamTextResponse, Error>

Streams text generation and tool execution using the language model.

This method performs streaming text generation, providing real-time access to response chunks as they are produced. It supports tool calling and execution in multiple steps, streaming intermediate results and handling tool interactions dynamically.

For non-streaming responses, use generate_text instead.

§Returns

A StreamTextResponse containing the stream of chunks and final conversation state.

§Errors

Returns an Error if the underlying language model fails to generate a response or if tool execution encounters an error.

§Examples

   use aisdk::{
       core::{LanguageModelRequest, LanguageModelStreamChunkType},
       providers::OpenAI,
   };
   use futures::StreamExt;

   async fn main() -> Result<(), Box<dyn std::error::Error>> {

       let openai = OpenAI::gpt_5();

       let mut stream = LanguageModelRequest::builder()
           .model(openai)
           .prompt("What is the meaning of life?")
           .build()
           .stream_text()
           .await?
           .stream;

        while let Some(chunk) = stream.next().await {
            if let LanguageModelStreamChunkType::Text(text) = chunk {
                println!("{}", text);
            }
        }
       
       Ok(())
   }