MiniMax H3 Max eight-second clip: $0.50 to $1.60 on Sume

An eight-second MiniMax H3 Max clip (minimax-h3-max) costs $0.50 at 480p, $0.80 at 768p and $1.60 at 1080p on Sume. Audio is native stereo and always produced, so there is no audio line on the bill.

Prices are the provider list per second times 1.25, rounded up to whole cents, using the rates in the repo's pricing table (fal list read by the team on 2026-09-08).

Rates

H3 Max is the row to use when you want 1080p from MiniMax. The catalog states that 1080p is a latent refinement from a native 768p render, not a native 1080p one, so the middle tier is the true native step up from 480p.

Limits that matter at eight seconds

The duration range is 5 to 15 seconds, so eight seconds is valid. The aspect ratios include 21:9, 16:9, 4:3, 1:1, 3:4 and 9:16. There is no generate_audio toggle: sending the field is not how you silence this model.

Reference-to-video is supported with up to nine images, three video clips and three audio clips, at most twelve references in total, and audio cannot be the only reference. Check the exact limits in the catalog constraints before you build on them.

• Text-to-video, first and last frame image-to-video, and reference-to-video.
• No seed, no size, and no provider.options on any /v1/videos model.
• Reserve at submit, capture on completion, usage.cost on the job.

Compare before you pick

At 768p, eight seconds of H3 Max is $0.80. The same eight seconds of Wan 3.0 at 720p is $1.00, and Wan has a 2 to 30 second range if you later need a longer take. Kling 3 at eight seconds with sound is $1.68. Pick on the features you need, not only the price: references, ratios and audio control differ by row.

Request

A minimal text-to-video call:

curl -X POST "https://api.sume.com/v1/videos" \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"minimax-h3-max","prompt":"Rain on a neon street, a cyclist passes","duration":8,"resolution":"768p","aspect_ratio":"16:9"}'