MiniMax Video Overview
OpenAI format and unified video interface for MiniMax-H3, organized by model series.
MiniMax Video
MiniMax video currently only documents the MiniMax-H3 model, providing three integration methods: the official-format /v1/video/generations route, the OpenAI-compatible /v1/videos route, and the unified video /v1/video/create and /v1/video/query.
All route sets share the same generation scenarios, determined automatically from the request parameters:
| Scenario | Parameters | Aspect Ratio |
|---|---|---|
| Text-to-video | prompt only | Default 16:9, cannot be adaptive |
| Image-to-video | image or images | Forced adaptive |
| Multi-modal reference video | metadata.reference_video / reference_audio | Default adaptive |
Image-to-video and multi-modal reference video are mutually exclusive and cannot be mixed.
Official Format
Use POST /v1/video/generations to create a task and GET /v1/video/generations/{task_id} to query, submitting a multimodal content array.
Create Video
POST /v1/video/generations, create tasks with a multimodal content array.
Query Task
GET /v1/video/generations/{task_id}, query official-format task status and results.
OpenAI Format
Use POST /v1/videos to submit, POST /v1/videos/{video_id}/remix to regenerate, and GET /v1/videos/{task_id} to query.
MiniMax Video Generation
POST /v1/videos, submit MiniMax-H3 tasks in OpenAI style.
Video Regeneration
POST /v1/videos/{video_id}/remix, regenerate 768P videos to 2K.
MiniMax Task Query
GET /v1/videos/{task_id}, OpenAI-style video object query.
Unified Video
Use POST /v1/video/create to create a task, POST /v1/video/remix to regenerate, and GET /v1/video/query?id= to query.
Create Video
POST /v1/video/create, submit MiniMax-H3 tasks in unified video format.
Video Regeneration
POST /v1/video/remix, regenerate 2K by task ID or source video.
Query Task
GET /v1/video/query, query status and results by task id.
Comparison
| Item | Official Format | OpenAI Format | Unified Video |
|---|---|---|---|
| Create | POST /v1/video/generations | POST /v1/videos | POST /v1/video/create |
| Regenerate | — | POST /v1/videos/{video_id}/remix | POST /v1/video/remix |
| Query | GET /v1/video/generations/{task_id} | GET /v1/videos/{task_id} | GET /v1/video/query?id= |
| Source task | — | Path parameter video_id | Request body task_id or metadata.base_video |
| Input structure | Multimodal content array | prompt + image and other fields | Same as OpenAI Format |
| Generation/query semantics | Identical across all three routes | Identical across both routes | Identical across both routes |