Struct GeminiEmbedding 

pub struct GeminiEmbedding { /* private fields */ }

Client for generating embeddings using Gemini embedding models.

Example

use gemini_client_api::gemini::embed::GeminiEmbedding;
use gemini_client_api::gemini::types::embedding::TaskType;

let embedder = GeminiEmbedding::new("YOUR_API_KEY", "gemini-embedding-001")
    .set_task_type(TaskType::RetrievalDocument);

let response = embedder.embed_text("Hello, world!").await.unwrap();
println!("Embedding dimension: {}", response.embedding().values().len());

Implementations

impl GeminiEmbedding

pub fn new(api_key: impl Into<String>, model: impl Into<String>) -> Self

Creates a new GeminiEmbedding client.

Arguments

• api_key - Your Gemini API key. Get one from Google AI studio.
• model - The embedding model to use (e.g., "gemini-embedding-001"). See embedding models.

Examples found in repository

examples/embedding.rs (line 10)

async fn main() {
    // 1. Create the Gemini Embedding client
    // Get your API key from https://aistudio.google.com/app/apikey
    let api_key = env::var("GEMINI_API_KEY").expect("GEMINI_API_KEY must be set");
    let embedder = GeminiEmbedding::new(api_key, "gemini-embedding-001")
        .set_task_type(TaskType::RetrievalDocument)
        // Optional: reduce dimension for Matryoshka Representation Learning
        .set_output_dimensionality(256);

    // 2. Generate embedding for a single text
    let prompt = "Rust is a blazing fast and memory-efficient systems programming language.";
    let response = embedder.embed_text(prompt).await.unwrap();

    // 3. Print the embedding information
    let embedding = response.embedding();
    println!("Embedding generated for: {:?}", prompt);
    println!("Total Dimensions: {}", embedding.dimension());
    println!(
        "First 5 values: {:?}",
        &embedding.values()[..usize::min(embedding.dimension(), 5)]
    );
}

pub fn new_with_client( api_key: impl Into<String>, model: impl Into<String>, client: Client, ) -> Self

Creates a new GeminiEmbedding client with a custom reqwest::Client.

Arguments

• api_key - Your Gemini API key.
• model - The embedding model to use.
• client - A custom reqwest::Client for making requests.

pub fn set_task_type(self, task_type: TaskType) -> Self

Sets the task type for the embedding.

The task type helps the model produce better embeddings tailored for the specific use case.

Examples found in repository

examples/embedding.rs (line 11)

async fn main() {
    // 1. Create the Gemini Embedding client
    // Get your API key from https://aistudio.google.com/app/apikey
    let api_key = env::var("GEMINI_API_KEY").expect("GEMINI_API_KEY must be set");
    let embedder = GeminiEmbedding::new(api_key, "gemini-embedding-001")
        .set_task_type(TaskType::RetrievalDocument)
        // Optional: reduce dimension for Matryoshka Representation Learning
        .set_output_dimensionality(256);

    // 2. Generate embedding for a single text
    let prompt = "Rust is a blazing fast and memory-efficient systems programming language.";
    let response = embedder.embed_text(prompt).await.unwrap();

    // 3. Print the embedding information
    let embedding = response.embedding();
    println!("Embedding generated for: {:?}", prompt);
    println!("Total Dimensions: {}", embedding.dimension());
    println!(
        "First 5 values: {:?}",
        &embedding.values()[..usize::min(embedding.dimension(), 5)]
    );
}

pub fn set_output_dimensionality(self, output_dimensionality: u32) -> Self

Sets the output dimensionality for the embedding.

Allows reducing the embedding dimension for storage/performance optimization via Matryoshka Representation Learning.

Examples found in repository

examples/embedding.rs (line 13)

async fn main() {
    // 1. Create the Gemini Embedding client
    // Get your API key from https://aistudio.google.com/app/apikey
    let api_key = env::var("GEMINI_API_KEY").expect("GEMINI_API_KEY must be set");
    let embedder = GeminiEmbedding::new(api_key, "gemini-embedding-001")
        .set_task_type(TaskType::RetrievalDocument)
        // Optional: reduce dimension for Matryoshka Representation Learning
        .set_output_dimensionality(256);

    // 2. Generate embedding for a single text
    let prompt = "Rust is a blazing fast and memory-efficient systems programming language.";
    let response = embedder.embed_text(prompt).await.unwrap();

    // 3. Print the embedding information
    let embedding = response.embedding();
    println!("Embedding generated for: {:?}", prompt);
    println!("Total Dimensions: {}", embedding.dimension());
    println!(
        "First 5 values: {:?}",
        &embedding.values()[..usize::min(embedding.dimension(), 5)]
    );
}

pub fn set_api_key(self, api_key: impl Into<String>) -> Self

pub fn set_model(self, model: impl Into<String>) -> Self

pub fn set_config(self, config: EmbedContentConfig) -> Self

Sets the full embedding configuration, replacing any previously set config.

pub async fn embed( &self, content: Vec<Part>, ) -> Result<EmbedContentResponse, GeminiResponseError>

Generates an embedding for the given content parts.

Arguments

• content - The content parts to embed (e.g., text, inline data).

Errors

Returns GeminiResponseError on network failure or API error.

pub async fn embed_text( &self, text: impl Into<String>, ) -> Result<EmbedContentResponse, GeminiResponseError>

Convenience method to generate an embedding for a single text string.

Arguments

• text - The text to embed.

Example

let embedder = GeminiEmbedding::new("YOUR_API_KEY", "gemini-embedding-001");
let response = embedder.embed_text("What is the meaning of life?").await.unwrap();
println!("Got {} dimensions", response.embedding().values().len());

Examples found in repository

examples/embedding.rs (line 17)

async fn main() {
    // 1. Create the Gemini Embedding client
    // Get your API key from https://aistudio.google.com/app/apikey
    let api_key = env::var("GEMINI_API_KEY").expect("GEMINI_API_KEY must be set");
    let embedder = GeminiEmbedding::new(api_key, "gemini-embedding-001")
        .set_task_type(TaskType::RetrievalDocument)
        // Optional: reduce dimension for Matryoshka Representation Learning
        .set_output_dimensionality(256);

    // 2. Generate embedding for a single text
    let prompt = "Rust is a blazing fast and memory-efficient systems programming language.";
    let response = embedder.embed_text(prompt).await.unwrap();

    // 3. Print the embedding information
    let embedding = response.embedding();
    println!("Embedding generated for: {:?}", prompt);
    println!("Total Dimensions: {}", embedding.dimension());
    println!(
        "First 5 values: {:?}",
        &embedding.values()[..usize::min(embedding.dimension(), 5)]
    );
}

Trait Implementations

impl Clone for GeminiEmbedding

fn clone(&self) -> GeminiEmbedding

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for GeminiEmbedding

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

Formats the value using the given formatter.