Tag
Jev-Omni is the first open-weight model to extend typed decisions to multimodal, supporting text, images, audio, and video, and directly returning the probability distribution of options without generating explanations.