mistral Mistral Releases Shieldstral, a 3B Open-Weight Model for On-Device Content Safety Shieldstral brings policy-adaptive, multimodal content moderation to a 3B open-weight model designed for local and edge deployment.
mistral Mistral Moderation API: What Its Documented Text Guardrails and Scores Actually Cover Mistral's public moderation materials describe a text-focused API with category-level scores, thresholds and endpoints for raw text and conversational content.
mistral Shieldstral Introduces Policy-Adaptive Multimodal Safety Classification in a 3B Model Shieldstral presents a compact approach to multimodal moderation, allowing safety criteria to be expressed through natural-language prompts rather than fixed labels.
