Ollama recently released version 0.32.9, which adds support for the NVIDIA Nemotron 3.5 Lightning model. This model is described as "an open 30B mixture-of-experts (MoE) model with 3B active parameters" specifically designed for the execution layer of always-on AI agents. The update enables users to run Nemotron 3.5 Lightning locally, facilitating development with agents utilizing harnesses such as OpenClaw and Hermes Agent. NVIDIA's NemoClaw open-source security and management stack supports these agent frameworks. The release notes also indicate the addition of the underlying Nemotron 3 architecture and improvements to the Muse Glimmer function calling parser.