Specifically meant to help image reasoning use scenarios and optimized to detect damaging multimodal (text and image) prompts and textual content responses to those prompts. as an alternative to swap it, generative AI https://youtu.be/aGubCnA5Qyk