Enum EmbeddingModelChoice 

pub enum EmbeddingModelChoice {
    BgeSmall,
    BgeBase,
    Nomic,
    GteLarge,
    OpenAILarge,
    OpenAISmall,
    OpenAIAda,
    Nvidia,
    Gemini,
    Mistral,
}

Supported embedding models for semantic search

Variants

BgeSmall

BGE-small-en-v1.5: Fast, 384-dim, ~78% accuracy (default)

BgeBase

BGE-base-en-v1.5: Balanced, 768-dim, ~85% accuracy

Nomic

Nomic-embed-text-v1.5: High accuracy, 768-dim, ~86% accuracy

GteLarge

GTE-large-en-v1.5: Best semantic depth, 1024-dim

OpenAILarge

OpenAI text-embedding-3-large: Highest quality, 3072-dim (requires OPENAI_API_KEY)

OpenAISmall

OpenAI text-embedding-3-small: Good quality, 1536-dim (requires OPENAI_API_KEY)

OpenAIAda

OpenAI text-embedding-ada-002: Legacy model, 1536-dim (requires OPENAI_API_KEY)

Nvidia

NVIDIA nv-embed-v1: High quality, remote embeddings (requires NVIDIA_API_KEY)

Gemini

Gemini text-embedding-004: Google AI embeddings, 768-dim (requires GOOGLE_API_KEY or GEMINI_API_KEY)

Mistral

Mistral mistral-embed: Mistral AI embeddings, 1024-dim (requires MISTRAL_API_KEY)

Implementations

impl EmbeddingModelChoice

pub fn is_openai(&self) -> bool

Check if this is an OpenAI model (requires OPENAI_API_KEY)

pub fn is_remote(&self) -> bool

Check if this is a remote/cloud model (not local fastembed)

pub fn to_fastembed_model(&self) -> EmbeddingModel

Get the fastembed EmbeddingModel enum value (only for local models)

Panics

Panics if called on an OpenAI model. Use is_openai() to check first.

pub fn name(&self) -> &'static str

Get human-readable model name

pub fn canonical_model_id(&self) -> &'static str

Get the canonical provider model identifier used for persisted metadata.

This is intended to match upstream provider IDs (OpenAI) and HuggingFace-style IDs (fastembed/ONNX) so that memories can record an embedding “identity” that other runtimes can select deterministically.

pub fn dimensions(&self) -> usize

Get embedding dimensions

impl EmbeddingModelChoice

pub fn from_dimension(dim: u32) -> Option<Self>

Infer the best embedding model from vector dimension stored in MV2 file.

This enables auto-detection: users don’t need to specify –query-embedding-model if the MV2 file has vectors. The dimension uniquely identifies the model family.

Dimension Mapping

• 384 → BGE-small (default local model)
• 768 → BGE-base (could also be Nomic, but same dimension works)
• 1024 → GTE-large
• 1536 → OpenAI small/ada
• 3072 → OpenAI large

Trait Implementations

impl Clone for EmbeddingModelChoice

fn clone(&self) -> EmbeddingModelChoice

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for EmbeddingModelChoice

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

Formats the value using the given formatter.

impl Default for EmbeddingModelChoice

fn default() -> EmbeddingModelChoice

Returns the “default value” for a type.

impl Display for EmbeddingModelChoice

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

Formats the value using the given formatter.

impl FromStr for EmbeddingModelChoice

type Err = Error

The associated error which can be returned from parsing.

fn from_str(s: &str) -> Result<Self>

Parses a string s to return a value of this type.

impl PartialEq for EmbeddingModelChoice

fn eq(&self, other: &EmbeddingModelChoice) -> 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 Copy for EmbeddingModelChoice

impl Eq for EmbeddingModelChoice

impl StructuralPartialEq for EmbeddingModelChoice