Voyage AI Provider

Embedding generation using Voyage AI's Embeddings API. Voyage provides state-of-the-art embeddings optimized for retrieval and RAG applications.

Requirements

• Voyage AI API key from dash.voyageai.com/api-keys

Configuration

{ok, State} = barrel_embed:init(#{
    embedder => {voyage, #{
        api_key => <<"...">>,                    % or use env var
        model => <<"voyage-3">>                  % default
    }}
}).

Options

Using Environment Variable

Set VOYAGE_API_KEY instead of passing in config:

export VOYAGE_API_KEY=pa-...

{ok, State} = barrel_embed:init(#{
    embedder => {voyage, #{}}  % uses env var
}).

Supported Models

Domain-Specific Models

Voyage offers specialized models for specific domains:

%% For code search
{ok, State} = barrel_embed:init(#{
    embedder => {voyage, #{model => <<"voyage-code-3">>}}
}).

%% For legal documents
{ok, State} = barrel_embed:init(#{
    embedder => {voyage, #{model => <<"voyage-law-2">>}}
}).

%% For financial documents
{ok, State} = barrel_embed:init(#{
    embedder => {voyage, #{model => <<"voyage-finance-2">>}}
}).

Example

%% Initialize
{ok, State} = barrel_embed:init(#{
    embedder => {voyage, #{
        api_key => <<"pa-...">>,
        model => <<"voyage-3">>
    }}
}).

%% Generate embeddings
{ok, Vec} = barrel_embed:embed(<<"Machine learning is fascinating">>, State).

%% Batch (more efficient for multiple texts)
{ok, Vecs} = barrel_embed:embed_batch([
    <<"Document about AI">>,
    <<"Document about databases">>,
    <<"Document about networking">>
], State).

Why Voyage AI?

• Top-ranked on MTEB benchmark for retrieval
• Optimized specifically for RAG and search
• Domain-specific models for code, law, finance
• OpenAI-compatible API format

Pricing

See voyageai.com/pricing for current rates.