Enable the custom embedding model for semantic indexing

  • Release version: Yokohama
  • Updated July 31, 2025
  • 1 minute to read

    • Add a new embedding model in the semantic indexing table so that the AI Search Retrieval Augmented Generation (RAG) application can use this model for semantic indexing.

    Before you begin

    Role required: admin

    Procedure

    1. Navigate to All, and then enter ais_semantic_embedding_model.list in the filter to go to the AI Search Semantic Embedding Models [sys_generative_ai_config] table.
    2. Select New.
    3. In the Name field, enter a unique name.
      For example, Azure OpenAI Large Text Embedding.
    4. In the Model Id field, enter a unique ID.
      An ID starts with a letter or number and may include letters, digits, periods (.), or hyphens (-) after the first character.
    5. In the One Extend Capability Definition field, select a BYOM capability definition that you created to set a provider for the embedding model.
    6. In the Model Config field, select an embedding model that is already configured.
    7. Select Active.
    8. Optional: If you want to configure batching for your embedding model, do these steps:
      Batching helps the embedding model to process multiple inputs at once. The minimum and maximum batch size values control how inputs are grouped and processed to call the embedding generation API.
      1. Select Batching Supported.
      2. In the Minimum Batch Size and Maximum Batch Size fields, enter the required values.
        For example, the minimum number of inputs allowed in a single batch is 4 and the maximum number of inputs that can be processed together in one batch is 16.
    9. In the Error Handler Extension Instance field, select an error handler instance.
      You create a scripted extension point to handle embedding generation errors that occur when custom embedding models generate semantic vectors. For more information, see Create an error handler extension point.
    10. Select Submit.

    What to do next

    Add your embedding model to the semantic index configuration to enable content ingestion with that model. For more information, see Configure semantic indexing settings for an indexed source.