AI Video APIOne model, two focused workflows
The API keeps the common text-to-video path minimal while still offering deterministic scene endpoints when a first or last frame matters.
Fast text-to-video and image-to-video API
MiniMax H3 Max API
Generate 480p or 768p video from text or controlled first and last frames.
MiniMax H3 Max supports focused text-to-video and first/last-frame image-to-video workflows through one asynchronous API. Create 5–15 second clips, choose six fixed aspect ratios for text generation, or provide one to two images for frame-guided output. Billing depends only on output duration and resolution; input images are free.
Read the official model and pricing documentation5–15s
Output duration
480p / 768p
Resolution
Output only
Billing
MiniMax
minimax-h3-maxH3 family sample
5–15s
Output duration
480p / 768p
Resolution
Output only
Billing
Model capabilities
Built for production API workflows, not one-off demos.
Use MiniMax H3 Max from the same platform surface as the rest of your video and image stack: API keys, credits, logs, webhooks, and docs stay consistent across providers.
Text-to-video by default
Omit generation_type to create video directly from a prompt. Choose 21:9, 16:9, 4:3, 1:1, 3:4, or 9:16 without sending reference media upstream.
First and last frame control
Set generation_type to first_and_last_frames and provide one image for the opening frame or two images for both endpoints. Frame mode automatically follows the supplied image composition.
Output-only pricing
Input images carry no credit charge. The estimate and final debit use the same per-second rate selected by output resolution, making cost predictable before submission.
Asynchronous production workflow
Submit through the unified video endpoint, receive a task ID, and collect the finished hosted video by callback or status polling using the same task shape as the rest of the API.
API workflow
Submit tasks, track progress, and return generated assets.
01
Choose text or frame mode
Leave generation_type out for text-to-video. For image-to-video, explicitly select first_and_last_frames and attach one or two public HTTP(S) image URLs.
02
Set output controls
Choose an integer duration from 5 to 15 seconds and either 480p or 768p. Text mode accepts six fixed ratios; frame mode derives its ratio from the images.
03
Submit and retrieve the result
POST to the unified generation endpoint, then use the returned task ID for polling or rely on a callback URL. Credits are based only on successful output duration and resolution.
POST /v1/videos/generations
curl -X POST https://api.aivideoapi.ai/v1/videos/generations \
-H "Authorization: Bearer sk-your-api-key" \
-H "Content-Type: application/json" \
-d '{
"model": "minimax-h3-max",
"input": {
"prompt": "A cinematic tracking shot through a rain-lit night market",
"duration": 5,
"resolution": "768p",
"aspect_ratio": "16:9"
}
}'What teams build
Short-form campaign videos
Product and e-commerce motion
Social video variations
Storyboard-to-video previews
Opening and ending frame transitions
Rapid concept visualization
Why choose MiniMax H3 Max
The MiniMax H3 Max API benefits that matter to developers.
Simple cost forecasting
Only resolution and output seconds affect credits. Ignored images in text mode and valid frame images never add input fees.
Unified integration surface
Existing task polling, callbacks, hosted outputs, authentication, and permission controls work without a separate integration stack.
Frequently asked questions
Answers about the MiniMax H3 Max API.
How do I request text-to-video?
Omit generation_type. Any string-array image_urls included in that request are ignored and are neither forwarded nor billed.
How many images can image-to-video use?
The first_and_last_frames mode requires one or two valid HTTP(S) image URLs. The first maps to the opening frame and the second, when present, maps to the ending frame.
What resolutions and durations are supported?
Choose 480p or 768p and an integer duration from 5 through 15 seconds. The defaults are 768p and 5 seconds.
Can I cancel a submitted task?
No. This model does not support task cancellation. Use polling or a callback to wait for the final task state.
Pricing and usage
Clear model options with shared credits.
480p output
12.7 credits/s
A 5-second output costs 63.5 credits
768p output
19.23 credits/s
Default; a 5-second output costs 96.15 credits
Input images
Free
One or two images in frame mode; no input-material charge
Start building with MiniMax H3 Max in AI Video API.
Create one API key, use one credit balance, and switch between video and image models without provider-specific plumbing.