Find any model, tool or page and jump to it.
MiniMax H3 is MiniMax's state-of-the-art multimodal video model. Designed for reference-driven work. Built to make the picture and the sound in one pass. Write a simple prompt and generate a clip that already has audio. Set a first frame and a last frame, and the shot travels between them. Attach up to nine reference images to hold a character, a product, or a look across shots. Attach reference videos to carry motion and editing rhythm, and reference audio to carry a voice. Score, dialogue, foley, and ambience arrive in time with the picture. Available in Visual Sandbox alongside other advanced AI video models, so you can generate, edit, and iterate in a single place. A powerful choice for short film, advertising, music videos, character animation, and social content.
| Tier | Price |
|---|---|
| 768p | $0.10/s |
| 2K | $0.17/s |
Billed per second of output video. 2K costs more per second than 768p. Visual Sandbox shows the exact price before you generate.