0/3 · 0/15s

0/3 audio files · 0/15s total · WAV or MP3

0/2500

MiniMax H3 Multimodal References

MiniMax H3 Model Capabilities

Three dedicated workflows

Switch between text, start/end frame, and multimodal reference generation without changing tools.

Native 2K, 5–15 seconds

Choose any whole-second duration from 5 to 15 seconds with native 2K output.

Multimodal reference control

Guide the result with up to 9 images, 3 videos, and 3 audio files while keeping their input order.