Released in late 2025, this closed-source model creates talking avatars. Its input consists of a static character image and an audio file with speech. The model analyzes the audio to map phonemes to facial movements (visemes), generating accurate lip-sync and natural facial animation without 3D modeling. It prioritizes high-fidelity lip synchronization over full-body movement, making it ideal for virtual presenters and educational videos. For similar capabilities, consider alternatives like VEED Fabric.