Large language models used by Now Assist in Document Intelligence

  • 릴리스 버전: Australia
  • 업데이트 날짜 2026년 03월 12일
  • 소요 시간: 3분

    • Now Assist in Document Intelligence uses large language models (LLMs) to perform generative AI and agentic AI capabilities.

    You can select the LLM when creating a use case. The LLM can also be selected for the AI agent to use when processing documents and images.

    Multimodal LLMs can support image mode, as they are capable of processing multiple types of inputs like text, sound, images, and more, to generate a text response. Processing may take longer for image inputs.

    주:
    The document and visual insights AI agent only uses multimodal LLMs. The text-only Now LLM Service is not a supported model for the AI agent.

    The following table lists the available LLMs for Now Assist in Document Intelligence.

    표 1. LLM options for Now Assist in Document Intelligence
    LLM Highlights
    Now Assist default Use the LLM selected as the default for all Now Assist skills. For more information, see .
    Now LLM Service - Large Text-only model used for AI-driven solutions to support natural language understanding, automation, and decision support.
    Now LLM Service - Small Text-only model used for enterprise AI applications by enhancing text-based automation and content generation within ServiceNow workflows.
    Google Cloud - Gemini Large Multimodal model with advanced reasoning and problem-solving capabilities.
    Google Cloud - Gemini Small Multimodal model with strong performance in summarization, rewriting, and content transformation.
    Azure OpenAI - GPT Large Multimodal model with a strong performance on tasks involving analysis, summarization, and multi-step problem solving.
    Azure OpenAI - GPT Small Multimodal model that is lightweight, high-efficiency, and effective for routine workflows with summarization, paraphrasing, and Q&A.
    Amazon Bedrock - Claude Large Multimodal model with strong context management for long documents and dialogs.
    Amazon Bedrock - Claude Small Multimodal model with lower latency and higher efficiency for real-time applications.

    For more information, see Large language models on the ServiceNow AI Platform.