MiniMax H3 (open weights) budget line — via apimodels.app

This model runs the budget MiniMax H3 line of apimodels.app, built on the open MiniMax H3 weights. It is a thin client: your inputs are sent to apimodels.app with your own apimodels.app API key, and the finished MP4 is returned here.

Not MiniMax’s official H3 Lite. MiniMax publishes its own H3 Lite as minimax/h3-lite. This model is a different H3 line that apimodels.app has offered since August 2026 (listed there as minimax-h3-lite): it does not take reference videos and takes minutes per clip.

• Model page: this H3 line on apimodels.app
• API reference (call it directly from your code): /docs/minimax-h3-lite
• Get an API key: apimodels.app/console/api-keys

What it does

One model, four ways to steer it:

• Text to video — a prompt alone.
• Reference images — up to 9 images of a character, outfit, product or scene to keep consistent. Refer to them in the prompt as Picture 1, Picture 2, …
• Reference audio — up to 3 clips (a voice line for lip-sync, or music to move to), together with at least one reference image. Refer to them as Audio 1, …
• First / last frame — anchor the opening frame, or both the opening and closing frames.

Output: 480p or 768p, any whole length from 1 to 15 seconds, 16:9, 9:16 or adaptive (follow the reference image).

Pricing

Billed by apimodels.app per second of output, from your apimodels.app balance:

Failed generations are not charged. Replicate additionally bills its normal CPU rate for the time this run waits for the video.

Things to know

• It is not a fast model. 480p · 3 s takes about 1–2 minutes; 768p · 10 s about 6 minutes; 15 s clips 8–11 minutes, longer when the queue is busy.
• No reference video. For motion transfer from a video, use MiniMax H3 on apimodels.app.
• First/last-frame mode cannot be combined with reference images or audio; a last frame needs a first frame.
• Success is not 100%. Safety filters can reject a prompt or an image, and the model can be temporarily unavailable. In both cases nothing is charged on apimodels.app.
• Uploaded files are sent inline with the request (about 90 MB total after encoding).

Call the same model from your code

curl https://api.apimodels.app/v1/video/generations \
  -H "Authorization: Bearer $APIMODELS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "minimax-h3-lite", "prompt": "A girl with headphones nods to the beat in a bright studio", "duration": 5, "resolution": "768p", "ratio": "16:9"}'

# then poll until data.state is "completed" and read data.resultUrls[0]
curl "https://api.apimodels.app/v1/video/generations?task_id=TASK_ID" \
  -H "Authorization: Bearer $APIMODELS_API_KEY"

Full parameter list, Python / Node examples and callbacks: API reference.

Credits

MiniMax H3 is developed by MiniMax. apimodels.app is an independent API platform and is not affiliated with MiniMax or Replicate.