Our latest moderation model with 128k context window and jailbreaking detection. Strong performance on complex multilingual data including long multi-turn conversations.
