A post-trained, throughput-optimized build of MiniMax H3 — clips in seconds rather than minutes (a 480P 5-second clip came back in 6-11 s in our tests), stronger prompt adherence, better aesthetics — billed per second: 480P at $0.015/s and 768P at $0.024/s, any whole length from 5 to 15 seconds. A prompt alone is text-to-video (six aspect ratios); a first frame (plus an optional last frame) is image-to-video, with the aspect ratio following the image. This line does not take multiple reference images, reference video or reference audio — use minimax-h3 or minimax-h3-lite for those.
| Resolution | Per second | 5s | 10s | 15s |
|---|---|---|---|---|
| 480P | $0.015 | $0.075 | $0.15 | $0.225 |
| 768P (default) | $0.024 | $0.12 | $0.24 | $0.36 |
Launch pricing. Billed for the seconds you ask for, any whole number from 5 to 15. First / last frames cost nothing extra. Charged only on success, refunded on failure.
Versus the other two H3 tiers: Turbo is the fastest of the three (seconds versus minutes) and slightly above Lite at 768P ($0.024 vs $0.02/s), but its only image input is a first / last frame; minimax-h3-lite takes up to 9 reference images and 3 reference audio clips; minimax-h3 adds reference video and 2K.
All requests carry the API key in the header:
Authorization: Bearer YOUR_API_KEY/api/v1/video/generations·GET/api/v1/video/generations?task_id=# Text-to-video (prompt only; aspect_ratio applies)
curl -X POST https://api.apimodels.app/v1/video/generations \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "minimax-h3-max-turbo",
"prompt": "A white kitten chases a butterfly across a sunlit garden, gentle tracking shot, soft afternoon light",
"duration": 5,
"resolution": "768p",
"aspect_ratio": "16:9"
}'
# Image-to-video: first frame (+ optional last frame). Aspect ratio follows the image.
curl -X POST https://api.apimodels.app/v1/video/generations \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "minimax-h3-max-turbo",
"prompt": "The camera slowly pulls back, clouds drift overhead, light shifts across the terrain",
"first_frame_url": "https://example.com/first.jpg",
"last_frame_url": "https://example.com/last.jpg",
"duration": 8,
"resolution": "768p"
}'
# Poll until completed (result files are kept 7 days)
curl "https://api.apimodels.app/v1/video/generations?task_id=TASK_ID" \
-H "Authorization: Bearer YOUR_API_KEY"| Field | Required | Type | Description |
|---|---|---|---|
| model | Yes | string | minimax-h3-max-turbo |
| prompt | Yes | string | Describe subject, action, camera motion and lighting; up to 5,000 characters. |
| duration | No | number | Any whole second from 5 to 15, default 5. Billed for this many seconds. |
| resolution | No | string | 480p ($0.015/s) or 768p ($0.024/s, default). |
| aspect_ratio | No | string | Text-to-video only: 21:9 / 16:9 (default) / 4:3 / 1:1 / 3:4 / 9:16. Ignored when a first frame is given. |
| first_frame_url | No | string | First frame (public URL or base64). image is also accepted. |
| last_frame_url | No | string | Last frame; requires first_frame_url. |
| prompt_expansion_mode | No | string | balanced (default, about 1 s) or quality (up to ~30 s for a richer rewrite of the prompt). |
| seed | No | number | Random seed; random when omitted. |
| callback_url | No | string | We POST the result to this URL on completion; omit it and poll the GET endpoint instead. |
Try it in the Playground
Per second on apimodels.app at launch pricing: 480P $0.015/s and 768P $0.024/s, any whole length from 5 to 15 seconds — a 768P 10-second clip is $0.24, a 480P 5-second clip $0.075. First and last frames cost nothing extra. Only successful generations are charged; failures are refunded.
POST /api/v1/video/generations with model minimax-h3-max-turbo and a prompt. Optional: duration 5-15 (default 5), resolution 480p or 768p (default 768p), aspect_ratio for text-to-video (21:9 / 16:9 / 4:3 / 1:1 / 3:4 / 9:16), first_frame_url and last_frame_url for image-to-video, prompt_expansion_mode balanced or quality, seed. Poll GET /api/v1/video/generations?task_id=xxx or pass callback_url. One API key covers every model on apimodels.app.
Only a first frame and an optional last frame. It does not take multiple reference images, reference video or reference audio — send those and the request returns a clear 400 instead of silently dropping them. For up to 9 reference images and 3 audio clips use minimax-h3-lite; for reference video and 2K use minimax-h3.
Turbo is a post-trained build of the same H3 family tuned for prompt adherence and speed — clips come back in seconds where Lite takes 5-10 minutes — at 768P $0.024/s versus $0.02/s on Lite, but its only image input is a first / last frame and its minimum length is 5 seconds. Lite takes up to 9 reference images and 3 reference audio clips and runs from 1 second. Both top out at 768P. Pick Turbo for speed on text-to-video and single-image animation; pick Lite when you need character or product references.