MiniMax H3 on Sume: "720p is not supported; use 768p" explained

Send resolution: "720p" to minimax-h3 or minimax-h3-max on Sume and you get 720p is not supported by model minimax-h3; use 768p. These two rows render natively at 480p and 768p, and 768p is a first-class size rather than a stand-in for 720p. Change the value to 768p and the same request is admitted. The H3 Max row also offers 1080p, which is a refinement of the native 768p render.

MiniMax's own announcement lists 2K and a 768p option for H3, which is why the 720p habit from other video models does not carry over.

What each side says

The vendor page and the Sume rows agree on 768p as an option, and differ on the top end. Sume's minimax-h3 row bills 480p and 768p and prices 2K and 4K only when requested; the minimax-h3-max row has no 2K or 4K at all.

Per-second rates at 768p

Sume bills the provider list times 1.25, per the Video Router page. At the list rates in Sume's catalog that works out as follows; confirm against the live catalog before budgeting.

• minimax-h3: 480p about $0.0625 per second, 768p about $0.075 per second.
• minimax-h3-max: 480p about $0.0625, 768p $0.10, 1080p $0.20 per second.
• A 10-second 768p clip is about $0.75 on minimax-h3 and $1.00 on minimax-h3-max.

The corrected request

Only the resolution changes. Keep the duration at 5 to 15 seconds.

curl -X POST https://api.sume.com/v1/video-router/generate \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: h3-768-001" \
  -d '{
    "model": "minimax-h3",
    "prompt": "A barista pours latte art, close shot, stereo cafe sound",
    "resolution": "768p",
    "duration": 10,
    "aspect_ratio": "9:16",
    "mode": "async"
  }'

Related limits

Native stereo audio is always produced on both rows, so there is no audio switch to set. See the video generation guide for the response shape and MiniMax's announcement for the vendor description.