Parameters for SuiteScript LLM Embedding in NetSuite 2026.1

TL;DR Opening

Starting in NetSuite 2026.1, the SuiteScript LLM embedding feature introduces new parameters for better control over generating semantic embeddings from text. These enhancements offer developers flexibility in configuring integration with the OCI Generative AI service.

What's New in SuiteScript LLM Parameters

In the 2026.1 release, several new parameters have been introduced in SuiteScript for refining how developers utilize the large language model (LLM) for embedding purposes. Here’s a breakdown of these parameters:

Available Parameters

• options.inputs (string[]): Required

  • An array of input strings for which you want to get embeddings.

• options.dimensions (number): Optional

  • Specifies the number of dimensions for the returned embeddings. Ranging from 1 to 1536, this allows developers to adjust the level of detail captured by the embeddings, with 1536 being the default for the supported model.
  • Context Note: The current embed model, Cohere Embed v4.0, outputs embeddings with 1536 dimensions, capturing more nuanced information than its predecessors.

• options.embedModelFamily (string): Optional

  • Defines the model family to be used. By default, the Cohere Embed model (cohere.embed-v4.0) is applied unless specified otherwise.

• options.ociConfig (Object): Optional

  • Configuration used for unlimited access through the OCI Generative AI service, allowing integration with Oracle Cloud accounts.

  • Configuration overrides from this parameter will supersede those set in the AI Preferences page within NetSuite.

  • Sub-parameters:

    • compartmentId (string): Identifies the compartment within the Oracle Cloud.
    • endpointId (string): Required for custom AI clusters.
    • fingerprint (string): Only NetSuite secrets are acceptable for this parameter.
    • privateKey (string): Needs to be a NetSuite secret as well.
    • tenancyId (string) & userId (string): Required OCIDs for tenancy and user identification.

• options.timeout (number): Optional

  • Defines the maximum waiting time for the LLM's response in milliseconds, defaulting at 30 seconds.

• options.truncate (string): Optional

  • Specifies how to truncate inputs that exceed 512 tokens using defined methods from the llm.Truncate enumeration.

Example Syntax

Here’s a simple representation of how to use the new embedding parameters in a script:

const response = llm.embed({
  inputs: ["Hello World!"],
  embedModelFamily: llm.EmbedModelFamily.COHERE_EMBED,
  dimensions: 512
});

Who This Affects

• Developers: Those building applications using the N/llm module will find these parameters crucial for optimizing their embedding strategies.
• Administrators: Configuration of OCI credentials for unlimited usage will require attention to these parameters.

Key Takeaways

• The SuiteScript LLM embedding parameters in NetSuite 2026.1 allow for flexible configuration with Oracle Cloud AI services.
• Developers can specify both the dimension size of embeddings and the model family used.
• OCI configuration details can be managed directly through SuiteScript to ensure unlimited access.
• Always consider the default settings to avoid unnecessary complexity in your embedded integration.