MiniMaxOct 8, 2026
MiniMax H3 Hyper
MiniMax H3 Hyper is a version of MiniMax’s H3 video model. This text-to-video variant generates 4- to 15-second clips at 768p with native audio.
$0.0305
per second of video
768p
max resolution
15s
longest clip, from 4s
Mode
Same prompt, other models
Every video model renders the same reference prompts, so their clips line up side by side. Hover a clip to play it.
Slow aerial drone shot gliding over a misty mountain valley at golden hour, sunlight piercing clouds onto a winding river, ancient pine forests on either side, ultra-smooth motion, professional color grading, atmospheric haze, 4K
Pricing
Billed per clip in USD, by length and resolution. Each resolution is drawn to scale.
| 768p | |
|---|---|
| 4s clip | $0.10 |
| 5s clip | $0.13 |
| 10s clip | $0.31 |
| 15s clip | $0.55 |
Every clip comes with audio. Quoted from POST /video/quote on Nov 19, 2026. Prices can change, so quote a job before you queue it.
Capabilities
3 modes, each with its own model ID.
Text to video
Takes prompt
minimax-h3-hyper-text-to-videoTakes prompt
Image to video
Takes prompt and image
minimax-h3-hyper-image-to-videoTakes prompt and image
Reference to video
Takes prompt, reference images and audio
minimax-h3-hyper-reference-to-videoTakes prompt, reference images and audio
AudioGenerated with every clip
Aspect ratios
16:921:94:31:13:49:16
Specs
- Model ID
minimax-h3-hyper-text-to-video- Mode
- Text to video
- Endpoint
POST /api/v1/video/queue- Resolutions
- 768p
- Lengths
- 4–15s, any whole second
- Aspect ratios
- 16:9, 21:9, 4:3, 1:1, 3:4, 9:16
- Audio
- Generated with every clip
- Inputs
- Prompt
- Prompt limit
- 7,000 characters
- Privacy
- Anonymized
- Weights
- Closed
- Added
- Oct 8, 2026
Quickstart
Video is asynchronous: queue a job, then poll for the clip.
- 1Quote
POST /video/quoteOptional. Prices the job in USD without charging for it. - 2Queue
POST /video/queueChecks and charges the job, then returns a queue_id at once. - 3Retrieve
POST /video/retrievePoll it with the queue_id until the video is ready.
# 1. Queue the job (returns a queue_id)
curl https://api.venice.ai/api/v1/video/queue \
-H "Authorization: Bearer $VENICE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "minimax-h3-hyper-text-to-video",
"prompt": "Slow aerial drone shot over a misty mountain valley at golden hour",
"duration": "5s",
"resolution": "768P",
"aspect_ratio": "16:9"
}'
# 2. Poll until the MP4 is ready (JSON with the status until then)
curl https://api.venice.ai/api/v1/video/retrieve \
-H "Authorization: Bearer $VENICE_API_KEY" \
-H "Content-Type: application/json" \
-d '{ "model": "minimax-h3-hyper-text-to-video", "queue_id": "<queue_id>" }' --output video.mp4
# If the queue response has a download_url, retrieve only reports the
# status: download the MP4 from that URL once it is COMPLETED.Base URL
https://api.venice.ai/api/v1Parameters
| Parameter | Values |
|---|---|
model | minimax-h3-hyper-text-to-video |
prompt | Up to 7,000 characters |
duration | 4–15s, any whole second |
resolution | 768P |
aspect_ratio | 16:9 | 21:9 | 4:3 | 1:1 | 3:4 | 9:16 |