manyframeMiniMax H3.
Bring a written direction and a purposeful media set into one video task. H3 has its own multimodal inputs, resolution choices and duration limits.
Available models and settings are shown in your studio. Review the quote before generating.
The route.
The right inputs.
Start from the creative job, then choose the model and settings that fit it.
Model ID MiniMax-H3
- Duration
- 4–15 seconds, in whole seconds; 5-second default
- Resolution
- 768P or 2K
- Reference limits
- Up to 9 images, 3 videos and 3 audio clips; total input video duration up to 15 seconds
- Frame mode
- Up to two frame images; cannot mix frame images with reference media
- Aspect ratios
- 21:9, 16:9, 4:3, 1:1, 3:4, 9:16; adaptive requires a visual input
Capabilities describe the Manyframe connector. Your studio shows the currently enabled route, supported settings and maximum charge.
Brief the role of every input.
Use the image to explain the subject, the video to demonstrate movement and the audio to supply sound direction. H3 requires a non-empty text item, so tell it how those inputs should contribute to the scene.
Choose either frame images or reference media. For an adaptive composition, include an image or video that provides the ratio. Review the quote with your input media included: H3's usage accounting can include reference images and input-video duration as well as the output.
Plan a reference-led shot ↗
Good questions. Clear answers.
Still curious? Browse all FAQs
Can I mix first-frame images and a reference video?
No. H3 separates frame inputs from reference media; mixing them is rejected.
Does H3 share the 2.3 duration and resolution settings?
No. H3 uses its own 4–15 second range and 768P/2K resolutions. Other Hailuo routes have different combinations.
Make the reference set work together.
Give each input a purpose and review the media-aware quote.