Three-shot vertical product preview: H3 Max, 6 s each, $1.80

A three-shot vertical product preview, each shot 6 seconds on MiniMax H3 Max, costs 3 x $0.60 = $1.80 at 768p on Sume. The 480p draft costs $1.125 and the 1080p final costs $3.60.

Each shot is its own job, so you can redo shot two without paying for one and three again. Prices come from the Sume price tool (read 2026-10-09).

The preview at three resolutions

A good order is draft all three at 480p, redo the failures at 480p, then render the winners at the final resolution. The last step is the only one you pay full price for.

One body per shot

Give each shot the same reference_image_urls (your screens or product shots) so the look matches, and change only the prompt. H3 Max takes up to 9 reference images, 3 reference videos and 3 audio files, 12 in all.

{
  "model": "minimax-h3-max",
  "prompt": "Shot 2 of 3. Thumb taps the checkout button, the screen slides to a confirmation",
  "reference_image_urls": [
    "https://example.com/screen-home.png",
    "https://example.com/screen-checkout.png"
  ],
  "resolution": "768p",
  "duration": 6,
  "aspect_ratio": "9:16",
  "mode": "async"
}

Longer previews

More shots or longer shots scale the price in a straight line. The table prices three, five and eight shots of 6 seconds at 768p.

Limits and gotchas

If a store or ad platform requires exact dimensions or durations, check that platform's own page; this post gives only the Sume side.

• Each shot is 5-15 seconds. A 4-second shot is refused.
• Screens rendered by a model may not match your real UI. Check the text on the generated screen before you ship.
• Audio is native and always on, so the three clips will carry three different soundtracks. Replace them with one track.
• Use one Idempotency-Key per shot, such as preview-v1-shot-2.