Mistral AI unveils policy-adaptive multimodal safety model Shieldstral

Mistral AI has released Shieldstral 1.0 3B, a new multimodal safety classifier designed for efficient content moderation. Unlike traditional models that predict fixed categories, Shieldstral adapts to natural language safety policies provided at inference time, allowing it to handle novel moderation criteria without retraining. This compact, open-weight model can process text, images, or both, and is optimized for low-resource environments. AI

IMPACT Enables more flexible and efficient content moderation, particularly for edge devices and novel safety policies.

RANK_REASON New model release from a frontier AI lab (Mistral AI). [lever_c_demoted from frontier_release: ic=1 ai=1.0]

Read on Hugging Face Trending Models →

• gpt-oss-safeguard-20b
• LlavaGuard
• Ministral-3-3B-Base-2512
• mistralai/Shieldstral-1.0-3B
• Nemotron-3.5-Content-Safety-4B
• Nemotron-3.5-Safety
• Pixtral
• Qwen3Guard
• ShieldGemma
• Shieldstral

• model release
• safety
• product

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →