Vultr
NVIDIA Nemotron 3.5 Content Safety is a 4B-parameter multimodal safety model built on Google's Gemma-3-4B-it foundation and fine-tuned by NVIDIA for guardrails workloads. Vultr announced Day Zero availability of the model, giving developers access to a compact small language model designed to moderate AI inputs and outputs across text and images in a single workflow. Key capabilities include 23 safety categories aligned with the Aegis v2 taxonomy, custom policy reasoning with explanations, multilingual moderation across 12 languages, and a 128K-token context window. The model supports both fixed taxonomy classification and customizable policy modes, letting teams align moderation with internal compliance or industry-specific governance requirements.