Mistral open-sources 3B multimodal safety classifier Shieldstral, supporting custom policies at inference without retraining
Original title:Solutions Introducing Shieldstral. August 4, 2026 By Mistral
AI summary
Mistral released Shieldstral, a 3B open-weight multimodal safety classifier, available for download under the Apache 2.0 license. It models content moderation as a policy-adaptive question-answering task. At inference, policies can be passed in as natural-language questions to return calibrated safety scores without retraining, uniformly covering moderation of text, images and prompt-response pairs, refusal detection and toxicity detection.
Key Takeaways
- Mistral released Shieldstral,a 3B open-weight multimodal safety classifier,available for download…
- 0 license.
- It models content moderation as a policy-adaptive question-answering task.
- At inference,policies can be passed in as natural-language questions to return calibrated safety…
Quick Facts
- Published
- Updated
- Source
- Mistral AI:News(网页)
- Category
- Model Releases
- AI score
- 63
原文给出了把内容审核改为推理时自适应问答的方法、数据构建思路和基准表现,适合想自建安全审核方案的开发者参考。
Related entities
Source & copyright
This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.
How to Cite
When citing this article, keep the title, date, and permalink, and credit Orange News as the source.
Orange News: Mistral open-sources 3B multimodal safety classifier Shieldstral, supporting custom policies at inference without retraining. August 4, 2026. https://new.nat6.net/en/news/cmses2cpo14kuro2ehdj6irxa/