Domino IQ server updates

The following Domino IQ features and enhancements are added in V14.5.1.

Server component updates

Domino IQ server can use NVIDIA CUDA-built AI inference server components that work with only Windows or Linux servers. Domino IQ server components can now run on both CPU and NVIDIA GPU enabled hardware built with the NVIDIA 12.9 Update 1 Toolkit, which specifies the compatible NVIDIA GPU drivers for Windows and Linux Domino servers.

For more information, see the CUDA Toolkit 12.9 Update 1 - Release Notes in the CUDA Release Notes 12.9 documentation.

New guard model option

Domino IQ now integrates guard models to enhance ethical and secure use of large language models (LLMs) in applications. Guard models, such as Meta’s Llamaguard 3, IBM’s Granite Guard 3, and Google’s ShieldGemma, act as a protective layer to prevent harmful content, data leaks, and prompt injection risks. The guard model automatically classifies content as risky or not, based on predefined guidelines.

For more information, see Understanding LLM guard models.

New server commands for Domino IQ statistics