Model ID
Capabilities
MiniMax H3 is a cinema-grade still-image model. Text-to-image produces photorealistic 2K frames from a prompt. Image-to-image re-renders a subject from up to nine references into a new scene, outfit, or style while keeping identity.