Struct EmbeddingClient 

pub struct EmbeddingClient { /* private fields */ }

Universal embedding client with provider cascade

Implementations

impl EmbeddingClient

pub async fn new(config: &EmbeddingConfig) -> Result<Self>

Create client with config-driven provider cascade

pub async fn from_legacy(config: &MlxConfig) -> Result<Self>

Create from legacy MlxConfig (backward compatibility)

pub async fn from_env() -> Result<Self>

Legacy constructor from env vars only

pub fn connected_to(&self) -> &str

Get which provider we’re connected to

pub fn required_dimension(&self) -> usize

Get required dimension

pub fn batch_limits(&self) -> (usize, usize)

Get the current runtime batch limits used for embedding requests.

pub fn clone_with_batch_limits( &self, max_chars: usize, max_items: usize, ) -> Self

Clone the client while overriding the runtime batch limits.

pub async fn embed(&mut self, text: &str) -> Result<Vec<f32>>

pub async fn embed_batch(&mut self, texts: &[String]) -> Result<Vec<Vec<f32>>>

Embed a batch of texts with intelligent batching to avoid OOM.

Large texts are chunked and only the first chunk is embedded. Batches are split to stay under max_batch_chars and max_batch_items. Failed chunks are retried individually with exponential backoff.

pub async fn rerank( &mut self, query: &str, documents: &[String], ) -> Result<Vec<(usize, f32)>>

Trait Implementations

impl Clone for EmbeddingClient

fn clone(&self) -> EmbeddingClient

Returns a duplicate of the value.

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

Performs copy-assignment from source.