Video
Minimax H3
Use it ↗MiniMax H3 generates video from text, first/last frames, or multimodal references. Native stereo audio; 5-15s clips up to 2K at 24fps.
MiniMax H3 turns text, first/last frames, or reference media into native 2K, 24fps clips with stereo audio (5-15s). Guide subjects and style with up to 9 reference images, motion with up to 3 videos, and optional audio. First/last frame mode locks shot endpoints; reference mode cannot mix with frames. Synced dialogue, SFX, and ambience generate in one pass. Ideal for brand films, product showcases, story beats, motion design, and game UI trailers.