Does MiniMax H3 support text-to-video?

Yes. Skip the trays and write a prompt. Set aspect ratio, 768P or 2K, and duration from 4 to 15 seconds, then cook. One still is the first frame; two stills are first and last. Three or more stills, or any video/audio refs, go to reference-to-video.

How should I tag references?

Cite files in upload order as Image 1, Video 1, Audio 1. Native stereo is always on. Audio refs need at least one image or video and are free.

What resolutions and durations does Soup.video support?

MP4 at 768P or 2K, from 4 to 15 seconds. Image-to-video follows the source still. Output is ✦ 8/s at 768P and ✦ 13/s at 2K. Video references add the same rate for input seconds. The first five images are free; extras are ✦ 4 each.

MiniMax H3 AI Text, Image & Reference-to-Video

A white kitten chases a butterfly across a sunlit garden. Gentle camera tracking, natural movement, soft afternoon light filtering through the leaves.

The camera slowly pulls back from the scene, revealing the full landscape as clouds drift overhead and light shifts across the terrain.

Image 1 is the female protagonist. Image 2 is her small dog. Keep the woman and dog consistent with their respective reference images while they walk together through a sunlit garden.

Why MiniMax H3?

Three modes, one form

Text-to-video, first/last-frame I2V, or multimodal R2V.

Native stereo

Dialogue, effects, and music rendered with the picture. Always on.

2K output

768P or 2K, 4–15 seconds, billed per second at MiniMax list.

Mixed references

Up to 9 stills, 3 clips, and 3 audio files in one cook.

Character lock

Tag Image 1, Video 1, Audio 1 to keep identity and motion.

Pricing

Official MiniMax list. Output is billed per generated second. A video reference adds the same rate for input seconds. Native stereo is included. First five images are free; extras are 4 credits each.

Credits per second

Res	Output	Input video
768P	✦ 8/s	✦ 8/s
2K	✦ 13/s	✦ 13/s

How MiniMax H3 works

MiniMax H3 (Hailuo 03) reads a prompt plus optional stills, clips, and audio in one context. Soup routes the cook: skip files for text-to-video, one or two stills for image-to-video, or mixed refs for reference-to-video.

Outputs are .mp4, at 768P or 2K, 4 to 15 seconds. Native stereo is always on — there is no audio toggle.

Capabilities

Text-to-video — no files. Write a prompt, pick a ratio, cook.
Image-to-video — one still is the first frame; two stills are first and last. Ratio follows the source.
Reference-to-video — 3–9 stills, or any video/audio refs. Tag them as Image 1, Video 1, Audio 1.

How to cook

1 Describe the scene, action, camera, and sound. Put spoken lines in the prompt.
2 Skip files for text-to-video, or attach stills, clips, and audio. Tag them in order.
3 Set 768P or 2K, aspect ratio, and duration. The run button shows the price.

Settings

prompt: Required. Up to 7,000 characters. Tag refs as Image 1, Video 1, Audio 1.
images: None for text-to-video. One still is the first frame; two stills are first and last. 3–9 stills route to reference-to-video. JPG, PNG, or WEBP. Up to 30MB each.
videos: Optional motion refs, up to 3 clips, combined length at most 15s. MP4 or MOV, up to 50MB each. Billed at the same per-second rate as output.
audio: Optional soundtrack refs, up to 3 clips, combined length at most 15s. MP3 or WAV, up to 15MB. Free. Need at least one image or video.
resolution: 768P (default) or 2K.
aspect_ratio: Text-to-video and reference-to-video: adaptive, 21:9, 16:9, 4:3, 1:1, 3:4, 9:16. Image-to-video follows the source still.
duration: Integer seconds, 4 through 15. Default 5.

FAQ

Does MiniMax H3 support text-to-video?: Yes. Skip the trays and write a prompt. Set aspect ratio, 768P or 2K, and duration from 4 to 15 seconds, then cook.
When does Soup use reference-to-video?: Three or more stills, or any video or audio reference. Tag files as Image 1, Video 1, Audio 1. One or two stills stay on image-to-video (first frame, or first and last).
Does MiniMax H3 generate audio?: Yes. Native stereo is always on — dialogue, effects, and music. Audio references are free and need at least one image or video.
How is MiniMax H3 priced?: Official MiniMax list: ✦ 8/s at 768P and ✦ 13/s at 2K. Video references add the same rate for input seconds. The first five images are free; extras are ✦ 4 each.