Module embed 

modo::embed

Text-to-vector embeddings via LLM provider APIs.

Provides:

• EmbeddingProvider — concrete wrapper for any embedding backend
• EmbeddingBackend — trait for custom embedding providers
• OpenAIEmbedding — OpenAI embedding provider
• GeminiEmbedding — Google Gemini embedding provider
• MistralEmbedding — Mistral embedding provider
• VoyageEmbedding — Voyage AI embedding provider
• OpenAIConfig / GeminiConfig / MistralConfig / VoyageConfig — provider configs
• to_f32_blob / from_f32_blob — vector ↔ blob conversion helpers
• test::InMemoryBackend — in-memory backend for unit tests (#[cfg(test)] or test-helpers)

Quick start

use modo::embed::{EmbeddingProvider, OpenAIEmbedding, OpenAIConfig};

let config = OpenAIConfig {
    api_key: "sk-...".into(),
    ..Default::default()
};
let client = reqwest::Client::new();
let embedder = EmbeddingProvider::new(
    OpenAIEmbedding::new(client, &config)?,
);

let blob = embedder.embed("hello world").await?;
// Store blob in libsql F32_BLOB column

Modules

test

Test helpers for the embedding module.

Structs

EmbeddingProvider

Concrete embedding provider — wraps any EmbeddingBackend.

GeminiConfig

Configuration for the Gemini embedding provider.

GeminiEmbedding

Google Gemini embedding provider.

MistralConfig

Configuration for the Mistral embedding provider.

MistralEmbedding

Mistral embedding provider.

OpenAIConfig

Configuration for the OpenAI embedding provider.

OpenAIEmbedding

OpenAI embedding provider.

VoyageConfig

Configuration for the Voyage AI embedding provider.

VoyageEmbedding

Voyage AI embedding provider.

Traits

EmbeddingBackend

Trait for embedding providers.

Functions

from_f32_blob

Decode a little-endian byte blob back to f32 values.

to_f32_blob

Encode an f32 slice to a little-endian byte blob suitable for libsql F32_BLOB columns.