MistralNew modelConfirmed
MistralCorrection: Shieldstral is a new model for developers to check text and images against their service's safety rules. Teams write the rules in plain language.
Mistral releases a specialist model for developers checking whether text or images break their service's safety rules.
Teams write those rules as plain-language questions and receive a score for each item.
- Who can use it
- Downloadable open weights under Apache 2.0. Hosted API availability and regional requirements are not specified by the source.
- When
- The August 4 announcement says the weights are available for download that day. An exact release time is not specified by the source.
Mistral AI official announcement Announced
