MiniMax

MiniMax H3

VIDEO

MiniMax-H3 is a video generation model on APIMart that supports text-to-video, image-to-video with first and last frame control, and multimodal reference-to-video using reference images, videos, and audio. It outputs 2K videos with audio and uses asynchronous task submission.

MiniMax-H3

PROVIDERSMiniMaxENDPOINTSvideo-generationCAPABILITIEStext_to_video · image_to_video · video_to_video

PRICING

Pricing is based on total video duration. Actual charges are based on the final output. The first 5 reference images are free; additional images are charged per image.

Resolution	Unit price
768p	$0.06855 / sec
2K / Other	$0.10973 / sec

Video cost: (generated video duration + reference video duration) × unit price
Reference image cost: max(number of reference images - 5, 0) × $0.02746

Parameter	Type	Default	Range / values	Description
`model` *	`enum`	`MiniMax-H3`	MiniMax-H3	Model name. Fixed value: MiniMax-H3.
`prompt` *	`string`	—	<= 7000 characters	Required video prompt. Describe the scene, subject, action, camera movement, and visual style.
`duration`	`integer`	`5`	4 ~ 15 seconds	Video duration in seconds. The upstream API only accepts integers from 4 to 15.
`resolution`	`enum`	`2K`	2K · 768P	Output video resolution. MiniMax-H3 only supports 2K.
`aspect_ratio`	`enum`	`16:9`	adaptive · 21:9 · 16:9 · 4:3 · 1:1 · 3:4 · 9:16	Output aspect ratio. For text-to-video, omit or use adaptive to fall back to 16:9; image-to-video follows the input image; multimodal reference mode defaults to adaptive.
`size`	`enum`	—	adaptive · 21:9 · 16:9 · 4:3 · 1:1 · 3:4 · 9:16	Compatibility alias for aspect_ratio. Use aspect_ratio as the preferred field when possible.
`ratio`	`enum`	—	adaptive · 21:9 · 16:9 · 4:3 · 1:1 · 3:4 · 9:16	Compatibility alias for aspect_ratio. Use aspect_ratio as the preferred field when possible.
`watermark`	`boolean`	`false`	—	Whether to add an AIGC watermark. The upstream default is false. Compatible alias: aigc_watermark.
`aigc_watermark`	`boolean`	—	—	Compatibility alias for watermark.
`first_frame_image`	`string`	—	—	First-frame image URL for image-to-video. Do not mix first/last frame fields with multimodal reference fields.
`last_frame_image`	`string`	—	—	Last-frame image URL for image-to-video. Can be used with first_frame_image for first/last frame control.
`image_urls`	`array`	—	<= 9 reference images	Reference image URL array for multimodal reference-to-video. Images in image_urls are always treated as reference_image, not first or last frames.
`video_urls`	`array`	—	<= 3 reference videos; MP4/MOV; each 2 ~ 15 s; total <= 15 s; file <= 50 MB	Reference video URL array for multimodal reference-to-video.
`audio_urls`	`array`	—	<= 3 reference audios; WAV/MP3; each 2 ~ 15 s; total <= 15 s; file <= 15 MB; must be used with image_urls or video_urls	Reference audio URL array. Cannot be used alone; it must be paired with reference images or reference videos.