Mistral's Shieldstral: 3B open-weights model for multimodal moderation

Mistral has released Shieldstral, a 3 billion-parameter open-weights multimodal safety classifier that performs comparably to models seven times its size by framing content moderation as a policy-adaptive question-answering task. Unlike traditional guardrail models with fixed harm categories, Shieldstral accepts plain-language safety policies at inference time without requiring retraining, providing calibrated safety scores for both text and images. The model is released under Apache 2.0 license and runs efficiently on a single 16GB GPU.