A tiger, rendered tender
A quiet animal portrait with a gentle handheld camera move
A tiger, rendered tender
MiniMax H3 is a multimodal video model for short-form text-to-video and image-to-video creation. This studio keeps the model decision out of the way and puts prompt, frame, duration, ratio, and resolution in one focused surface.
Start from language when the idea is still open. Start from a first frame when composition, product shape, or character identity already matters. Both paths use the same transparent credit quote before a paid render begins.
A quiet animal portrait with a gentle handheld camera move
A surreal flood rolling through a sun-washed Italian village
A fashion film in a red room with tactile paper and soft light
A flooded street, a white sofa, a tiny absurdist moment
Every plan uses the same H3 engine. Credits show the exact cost before a task leaves the studio.
900 credits
112 sec 768P / 69 sec 2K
2,000 credits
250 sec 768P / 153 sec 2K
4,200 credits
525 sec 768P / 323 sec 2K
A multimodal video model for creating short moving images from text and first-frame images. H3 supports 768P and 2K output with 4–15 second durations.
No. You can explore curated examples and the full generator before signing in. A render requires a Google account and purchased credits.
If there is no playable result, the full credit quote is returned automatically. You can retry with the original prompt.
Private videos are kept for 30 days. Public videos remain available until you remove them or they are taken down after review.