MiniMax API · Source video
Provider source (opens in a new tab)- API model ID
MiniMax-H3- Access
- POST https://api.minimax.io/v2/video_regeneration with Bearer API key and Content-Type: application/json. Query GET https://api.minimax.io/v2/query/video_generation/{task_id}.
- Regions
- International API; inference residency not verified
- Context / input limit
- Not collected
- Maximum input tokens
- Not collected
- Maximum output
- Not collected
- Reasoning
- Not collected
- Tool calling
- Not collected
- Structured output
- Not collected
- Streaming
- Not collected
- Batch processing
- Not collected
- Fine-tuning
- Not collected
- Knowledge cutoff
- Not collected
- Release date
- Not collected
- Parameters
- Not collected
- Architecture
- In-context regeneration using the H3 base model plus the original multimodal contextSource for Architecture (opens in a new tab)
- Languages
- Not collected
- Context caching
- Not collected
- API aliases
- MiniMax-H3Source for API aliases (opens in a new tab)
- Input modalities
- Text, Image, Video, AudioSource for Input modalities (opens in a new tab)
- Request body limit
- 64 MB total request bodySource for Request body limit (opens in a new tab)
- Input media transport
- Public URLs, mm_file://{file_id}, or media-specific Base64 data URIs. URLs/platform files are recommended for large inputs; Base64 adds about 33% overhead.Source for Input media transport (opens in a new tab)
- Asynchronous generation
- SupportedSource for Asynchronous generation (opens in a new tab)
- Task ID lifetime
- Only tasks from the last seven days can be queried: [T−7d, T), with T in UTC seconds.Source for Task ID lifetime (opens in a new tab)
- Task callbacks
- Callback URL must echo the challenge within three seconds. Verified callbacks receive queued, running, succeeded, failed or cancelled updates.Source for Task callbacks (opens in a new tab)
- Cancellation and task deletion
- Queued tasks can be cancelled without charge. Running tasks cannot be cancelled. Succeeded/failed task records can be deleted; cancelled records cannot use this action.Source for Cancellation and task deletion (opens in a new tab)
- rateLimit
- Not collected
- Published quota scope
- Published family table: Video Generation V2 / MiniMax-H3, 300 RPM and 30 inflight tasks. Component-specific applicability remains unverified.Source for Published quota scope (opens in a new tab)
- Output modalities
- VideoSource for Output modalities (opens in a new tab)
- Maximum prompt characters
- 40,000Source for Maximum prompt characters (opens in a new tab)
- Output resolutions
- 2KSource for Output resolutions (opens in a new tab)
- Video input duration
- 107–362 frames in increments of 17, at 24 fps (provider describes approximately 4–15 seconds)Source for Video input duration (opens in a new tab)
- Required source video format
- Exactly one base_video with an audio track; 24 fps; both dimensions divisible by 32; area 589,824–1,032,192 pixels; 107–362 frames in steps of 17.Source for Required source video format (opens in a new tab)
- Original generation context
- Supply exactly the same final prompt and reference images/videos/audio used to generate the 768P source. Mismatched context may prevent the expected result.Source for Original generation context (opens in a new tab)
- Output link expiry
- Video download URLs expire; query the task again for a fresh URL within the seven-day task-query window. Exact URL lifetime is not published here.Source for Output link expiry (opens in a new tab)
- Input modes
- Choose exactly one: source_task_id, or content containing exactly one role=base_video. Supplying both or neither is invalid.Source for Input modes (opens in a new tab)
- Input image formats
- JPG, JPEG, PNG, WEBP, HEIC, HEIFSource for Input image formats (opens in a new tab)
- Image input limits
- ≤30 MB per image; each side 256–5,760 pixels; aspect ratio 0.4–2.5. At most one first frame, one last frame, or nine reference images.Source for Image input limits (opens in a new tab)
- Reference images
- One first frame, one last frame, or up to nine reference images. First/last-frame roles cannot be mixed with reference roles.Source for Reference images (opens in a new tab)
- Input video limits
- At most three reference clips; each 2–15 seconds, total ≤15 seconds; ≤50 MB per clip. MP4/MOV; H.264/AVC or H.265/HEVC with AAC/MP3 audio; 256–5,760 pixels per side, aspect ratio 0.4–2.5, input frame rate 23.976–60 fps.Source for Input video limits (opens in a new tab)
- Reference audio limits
- At most three WAV/MP3 reference clips; each 2–15 seconds, total ≤15 seconds; ≤15 MB per clip.Source for Reference audio limits (opens in a new tab)
- Required request fields
- model=MiniMax-H3, resolution=2K and content with exactly one base_video plus unchanged original generation inputs; no source_task_id.Source for Required request fields (opens in a new tab)
- Visible watermark
- aigc_watermark=false by default; true adds an AIGC label.Source for Visible watermark (opens in a new tab)
Shared selector; the task route identifies this component.
A JSON task envelope or structured prompt text does not establish caller-defined JSON Schema constrained generation.
The reviewed contract describes asynchronous task completion, not incremental output streaming.
Asynchronous tasks, lists and callbacks are not a documented Batch Inference API.
Public H3 Base training code/weights do not establish fine-tuning of this managed component.
System is not yet open-sourced; this is not proof of distinct private base weights.
No prompt-cache API, eligibility, cache lifetime or billing rate is documented for this component. Internal tensor/activation caching in H3 Base is not endpoint prompt caching.
Shared API selector for the H3 family; task route/type identifies this component. Not a separate base checkpoint ID.
Original context plus an H3-compatible base video, or a eligible source task ID.
Returns a task_id; retrieve the result with the shared query endpoint. No inference call performed.
Deleting a task record does not establish erasure of all provider-held inputs or outputs.
The quota table gives Video Generation V2 / MiniMax-H3 300 RPM and 30 inflight tasks, but does not expressly scope that pool to IR or regeneration. No inherited numeric component quota.
Per text item for direct-source mode. Use the final prompt sent to the original generator, after any Context IR processing.
The frame-count lattice is authoritative; do not replace it with arbitrary integer-second durations.
Prices retain their published currencies and billing units below.
| Category | Usage | Rate | Unit |
|---|---|---|---|
| Video regeneration | 768P → 2K output1 | $0.05 | second of regenerated output |
| Video regeneration | Original input video2 | $0.05 | second of original input video |
| Video regeneration | Images beyond the first five3 | $0.025 | additional input image |
| Video regeneration | Audio input4 | $0.00 | audio input |
Rate conditions
- Standalone regeneration output charge. Original 768P generation is billed separately; input charges below are additional.
- Published input-video duration charge from the original 768P task. Do not silently charge the base video a second time as a reference video; exact metering depends on reported usage.
- The first five original input images are free; each additional image is USD 0.025.
- Published as free input audio for regeneration only. Output video and applicable original input-video/image charges remain payable.
Data & enterprise terms
Task records are queryable for seven days. This does not establish input retention, training use, inference location or complete provider-side deletion. No authenticated request was performed.