At a Glance · AI-generated

Mistral AI released Shieldstral 1.0, a 3-billion-parameter open-weights safety classifier designed to moderate AI model outputs. The lightweight model, introduced August 5, matches the performance of safety classifiers seven times its size while remaining policy-adaptive and capable of processing both text and images. The open-source release aims to provide developers with an efficient, accessible tool for content moderation across various AI applications.