Choose a workflow, add your direction, and preview the exact credit cost before generating.
Your MiniMax H3 result will appear here.
Create Native-Audio Video with MiniMax H3
Start from a prompt, animate exact first and last frames, or combine image, video, and audio references in one generation. MiniMax H3 keeps visual identity, motion, camera language, and sound working together in clips up to 15 seconds and 2K.
One multimodal context
Direct characters, products, motion, camera behavior, dialogue, music, and ambience together instead of stitching separate tools.
Exact frame control
Animate a first frame or define both opening and ending frames for controlled visual transitions.
Native stereo sound
Generate sound with the video and optionally use audio references for voice, timing, rhythm, or atmosphere.
Questions fréquemment posées
What is MiniMax H3?
MiniMax H3, also known as Hailuo 03, is a multimodal video model that understands text, images, video, and audio together. It creates video with native stereo sound at 768P or 2K.
Which generation modes are available?
Text mode creates from a prompt. Frames mode animates a required first frame and an optional last frame. References mode combines multiple images, videos, and audio files to guide identity, style, motion, camera work, voice, or rhythm.
How many credits does MiniMax H3 cost?
768P costs 16 credits per second and 2K costs 26 credits per second. Output duration and total reference-video duration are both billed at that rate. The first 5 input images are free; images 6–9 cost 8 credits each. Audio references are free.
What files can I upload?
Images support JPG, PNG, and WEBP up to 30MB. Reference videos support MP4 and MOV up to 50MB. Reference audio supports MP3 and WAV up to 15MB. References mode accepts up to 9 images, 3 videos, and 3 audio files.
Why does a reference video increase the cost?
The provider bills input video at the same per-second rate as generated video. Viw AI reads the uploaded video's real duration on the server and includes it in the displayed and final credit calculation.
What duration, resolution, and aspect ratios are supported?
Generate 4–15 second clips at 768P or 2K. Text and reference modes support 21:9, 16:9, 4:3, 1:1, 3:4, and 9:16; reference mode also supports Adaptive. Frames mode follows the uploaded frame composition.
Are failed generations refunded?
Yes. Credits are reserved when the task is submitted and automatically refunded if the provider rejects the request or the generation finishes with a failure.
