Introducing Shieldstral
If your organisation moderates user-generated content (forums, LMS submissions, chat), Shieldstral gives you a free, self-hostable safety filter you can point at your own content rules written in plain English — no ML expertise or model retraining needed, and it runs on a single consumer-grade GPU.
What changed
Source summary
Mistral released Shieldstral, a 3B-parameter open-weights safety classifier for content moderation across text and images. Unlike fixed-category guardrail models, it evaluates content against plain-language policies supplied at inference time without retraining, runs on a single 16GB GPU, and is released under Apache 2.0.
Why it matters to your work
Practical interpretation
If your organisation moderates user-generated content (forums, LMS submissions, chat), Shieldstral gives you a free, self-hostable safety filter you can point at your own content rules written in plain English — no ML expertise or model retraining needed, and it runs on a single consumer-grade GPU.
Source & availability
This is a summary of a published provider announcement, with practical context from SCALA’s feed.
Plan, region, workspace and rollout conditions may affect access in Australia. Check the original source for current availability.
mistral.aiThis update is primarily for technical teams.