Voyage AI

Voyage AI provides embedding models for semantic search and related tasks. seekdb's VoyageaiEmbeddingFunction lets you call Voyage AI from pyseekdb so your collections can use Voyage embeddings for RAG, agent memory, and other applications that depend on semantic understanding.

tip

Using Voyage AI service requires you to follow Voyage AI's pricing rules and may incur corresponding fees. Before proceeding, please visit their official website or refer to relevant documentation to confirm and accept their pricing standards. If you do not agree, please do not proceed.

Dependencies and authentication

• Install the pyseekdb package.
• A valid Voyage AI API key for authentication.

Example: create a Voyage AI embedding function

Import and initialize VoyageaiEmbeddingFunction. Configuration is often provided via environment variables for easier deployment.

• Basic usage

  Use the default environment variable VOYAGE_API_KEY and specify the model. The voyage-4-large model supports long context and high-accuracy retrieval.

  from pyseekdb.utils.embedding_functions import VoyageaiEmbeddingFunction
  
  ef = VoyageaiEmbeddingFunction(
      model_name="voyage-4-large"
  )

• RAG-oriented usage

  For RAG, set input_type to distinguish document indexing from query encoding.

  from pyseekdb.utils.embedding_functions import VoyageaiEmbeddingFunction
  
  ef = VoyageaiEmbeddingFunction(
      model_name="voyage-4-large",
      input_type="document"  # "document" or "query"; use "document" for collection documents
  )

• Custom configuration

  With models such as voyage-4-large, you can reduce the output dimension (for example, from 1024 to 512) to lower storage and speed up search while preserving accuracy.

  from pyseekdb.utils.embedding_functions import VoyageaiEmbeddingFunction
  
  ef = VoyageaiEmbeddingFunction(
      model_name="voyage-4-large",
      output_dimension=512  # Reduce dimension for storage and compute efficiency
  )

Parameters:

• model_name: Voyage AI model name (for example, voyage-3-large, voyage-4-large).
• api_key_env: Environment variable name for the API key (default: VOYAGE_API_KEY).
• input_type: Optional hint for retrieval: use "document" for documents being indexed and "query" for query texts.