Video ● Operational Service status →

MiniMax H3

⧉

minimax-h3

MiniMax H3 (Hailuo 03) video model: 1080P HD output, native audio always generated, 5-15 seconds in one-second steps, six aspect ratios. Text-to-video without a reference image, image-to-video with one (used as the first frame). Asynchronous: submit, then poll the job id for a re-hosted mp4 link (kept for 15 days). Billed per second; failed jobs are refunded in full.

Text-to-videoImage-to-video1080PNative audio

Get API Key → API Docs

Service Highlights

1080P HD output
Native audio always generated, no extra parameter
5-15 seconds, selectable per second
Billed per second, failed jobs refunded in full

Measured on real calls

Customer requests over the last 30 days

Success rate92%

Median render time per clip4.6 min

90% of jobs finish within13.6 min

Pricing

Per second$0.036

Shortest 5s$0.180

Longest 15s$0.540

Billed per second = rate x duration (5–15s, any integer). 1080P cost the same. The hold at submit equals the final settle exactly; failures are refunded in full.

View detailed pricing

Live Test · Playground

Try out MiniMax H3 right here (available after login).

Input

Prompt

"Beneath the Walls of Troy" - an epic action short. @Image1 is a Greek hero in golden armour and a crimson cloak with a long spear and a lion shield; @Image2 is a Trojan champion in black armour and a blue cloak with a curved sword and a horse shield; the setting follows the walls of Troy and the sunset battlefield of @Image3. Keep each warrior's armour, helmet, cloak colour and shield emblem exactly as in their own reference - never swap them.
Opening: wind drives the sand as the two face off twenty paces apart, sharp eyes behind their helmets.
Middle: they charge at once; the spear strikes the horse shield in a burst of sparks; rapid slashes and blocks up close as the camera orbits them 360 degrees.
Climax: the setting sun sits right between them as they clash mid-air in extreme slow motion, sparks, sand and the crimson cloak hanging in the air.
Ending: both lower their weapons and salute each other.
Cinematic action-commercial look, 24fps film grain, fluid and weighty fighting; war drums, clashing metal and an epic score with the big hits on the beat. No blood or wounds, no extra characters, no on-screen text or watermarks, no deformed weapons or limbs.

0 / 2000 bytes

Duration

Resolution

Aspect ratio

Reference image role

Reference images0 / 5

Click to upload or drag files here jpg / png / webp / gif, up to 16 MB each or paste links

With multiple references, address them in the prompt as @Image1 / @Image2, e.g. "the character from @Image1 stands in the scene from @Image2". Without tokens the model decides on its own.

Reference audio0 / 3

Click to upload or drag files here mp3 / wav / m4a, up to 15 MB and 30 s each or paste links

Needs a reference image or video as well. Address it as @Audio1 in the prompt.

Uploaded references are kept for 7 days.

Rendering usually takes 5-20 min; when the backend is busy it can exceed 60 min (no result after 120 min counts as a timeout, fully refunded). Poll the job id rather than resubmitting - each resubmission is billed separately. Billed per second, fully refunded on failure. Prompt limit 2000 bytes (about 666 Chinese characters).

Output

video

Results will appear here after running.

🕑 Generated images and videos are kept for 15 days before 2026-10-29 09:49:40 UTC; from that time, they are kept for 7 days and existing files older than 7 days are removed.

About MiniMax H3

MiniMax H3 (Hailuo 03) is the highest-resolution video model here: 1080P output with native audio always on. Duration is any integer 5-15 seconds across six aspect ratios; omit the reference image for text-to-video, include one (first frame by default) for image-to-video. Async job: POST /v1/videos/generations returns HTTP 202 and a job id immediately; poll GET /v1/videos/jobs/{id} until status=succeeded and read data[0].url. That link is an mp4 re-hosted by us — it is kept for 15 days (far longer than the upstream CDN link), so download it or copy it to your own storage before then.

Use cases

High-definition output

1920x1088, with the best per-frame detail and lighting of any video model on the station.

Clips with sound, no audio pass

Native audio is always generated with no extra parameters, removing a whole production step.

Full camera choreography

It handled a tracking shot into a whip-tilt into a low-angle finish, so it suits work with a real shot list.

Image-to-video

With the reference as the first frame, subject and composition stay stable — good for turning concept art into motion.

How to choose

Pick it when you want resolution, native audio, or full camera choreography. For more than 15 seconds use seedance-2.5; to turn the soundtrack off, or to drive from a source video, use the Seedance models. It also renders the slowest here (about 11 minutes for a 15-second clip), so use seedance-2.0-fast when speed matters.

FAQ

Can I turn the audio off?

No. This model always generates native audio, and audio=false is explicitly rejected rather than silently ignored. For silent clips use the Seedance models with audio=false.

Why does it take so long?

1080P with native audio is simply heavy: roughly 4 minutes for 5 seconds and 11 minutes for 15. It is an async job, so you hold a job id and poll — no connection is kept open.

How does image-to-video work?

Pass an image (public URL, data: URI or base64). By default it becomes the first frame, so the clip keeps the reference subject and framing. A second image becomes the last frame; set image_role=reference to switch to subject/style reference mode instead. The first/last-frame and reference modes are mutually exclusive.

How is it billed, and am I charged on failure?

Billed per second: the hold placed at submit is computed from the duration you asked for, and the settle is the exact same number. A failed or timed-out render refunds the whole hold, so you never pay for a clip you did not receive.