Storia in 6 fonti

Mistral's open model Shieldstral matches much larger safety models at a fraction of the size

Mistral's new 3B Shieldstral model checks AI inputs and outputs for safety violations using natural language yes-or-no questions instead of fixed categories. It matches models seven times its size in some benchmarks. Operators can set their own criteria at runtime rather than rely on a third party's category system, and the model can run locally.

Raccontata da

Confronto fonti

6 prospettive sulla stessa storia

AI · summaries

the-decoder.comStai leggendo1 mese fa

Mistral's open model Shieldstral matches much larger safety models at a fraction of the size

Mistral's 3B Shieldstral ties 20B GPT-OSS-Safeguard on safety benchmarks (F1 84.9%) via single-token answers. Operators define custom filtering rules at runtime per-app without retraining, reducing classifier overhead and costly false positives on production.

originale

Timeline cronologica

martedì 4 agosto 2026·mistral.ai
Introducing Shieldstral. | Mistral AI
Shieldstral introduces a 3B open-weights multimodal safety classifier that outperforms models up to 7x its size.
mercoledì 5 agosto 2026·dev.to
Shieldstral Introduces Policy-Adaptive Multimodal Safety Classification in a 3B Model
Shieldstral is a 3B-parameter policy-adaptive multimodal safety classifier designed to assess text...

Mistral's open model Shieldstral matches much larger safety models at a fraction of the size

Confronto fonti

Mistral's open model Shieldstral matches much larger safety models at a fraction of the size

Timeline cronologica

Introducing Shieldstral. | Mistral AI

Shieldstral Introduces Policy-Adaptive Multimodal Safety Classification in a 3B Model

Introducing Shieldstral. | Mistral AI

Mistral AI Releases Shieldstral 1.0 3B: An Open-Weights Policy-Adaptive Multimodal Safety Classifier Matching…

Mistral introduces Shieldstral to provide lightweight policy-aware moderation for AI models - SiliconANGLE

Mistral veröffentlicht Modell mit neuer Technik für Sicherheitsklassifikationen

Shieldstral Introduces Policy-Adaptive Multimodal Safety Classification in a 3B Model

Mistral introduces Shieldstral to provide lightweight policy-aware moderation for AI models - SiliconANGLE

Mistral's open model Shieldstral matches much larger safety models at a fraction of the size

Mistral veröffentlicht Modell mit neuer Technik für Sicherheitsklassifikationen

Mistral AI Releases Shieldstral 1.0 3B: An Open-Weights Policy-Adaptive Multimodal Safety Classifier Matching Models 7× Its Size