Get 50% Off MiniMax H3 and Other Selected Video Models

MiniMax H3 Text to Video

Type what you want to see. MiniMax H3 text to video gives you a finished 4-to-15-second clip — picture, dialogue and soundtrack in one MP4, at 768P or 2K. No footage, no editing app.

13 real clips · Press Try this to load one

A rapper falling between skyscrapers, shot from below

11s · 16:9 · 768P

Four scenes, one character

What one prompt gets you

What you get from one prompt
with MiniMax H3 text to video

Most text to video attempts die the same way. You write one sentence, the model fills in everything you did not say, and what comes back is nothing like the thing in your head.

You wrote

A woman walks down a street.

The model filled in the rest

time of day?
wardrobe?
camera move?
weather?
sound?
lens?

A figure walking into a neon-lit street at dusk, the camera pushing in slowly — What came back — picture, rain and the spoken line in one MP4

MiniMax H3 text to video gives you room to say more — up to 7,000 characters — and returns any whole number of seconds from 4 to 15, at 24 fps, in 768P or 2K, with the dialogue, effects and music already in the same file as the picture. No source image, no storyboard, no editing app.

Characters you can write: 7,000
Any whole second: 4–15 s
Frame rate: 24 fps
Resolution: 768P / 2K
Picture, dialogue, effects and music: 1 MP4

No source image
No storyboard
No editing app

Shot ladder

From one line to a shot: three ways to write
a MiniMax H3 prompt

One idea, three ways, each with the clip it produced — run all three and see what MiniMax H3 text to video is really reading.

Level 1
6 words
A woman walks down a street.
Drifts
Subject drifts between frames. The camera picks its own move.
Level 2
30 words
30 words
A woman in a red coat walks down a wet Tokyo alley at night, neon reflected in puddles, slow push-in with small amplitude, rain on metal and distant traffic.
Usable
The camera does what you asked and the rain is actually there.
The structure the model was trained on
Level 3
Official fields
Official fields
integrated_multimodal_description
[Shot 1] Red coat, wet Tokyo alley, slow push-in.
overall_soundscape
Rain on metal, distant traffic, footsteps.
non_diegetic_music
Low synth pad, slow pulse, under the dialogue.
Deliverable
Dialogue sits in front of the music because you said so.

More worked examples in prompt ideas — every one carries its full structured prompt.

Prompt builder

The MiniMax H3 text to video prompt structure, field by field

Fill three boxes, watch the prompt assemble itself, send it straight to the generator.

You do not have to guess at this. H3 was trained against a three-field structure, and once you write in it the model stops filling in the blanks for you.

Order locked

Where the words go

integrated_multimodal_description — the picture

Decides whether you get your shot or somebody else's.

Framing
Subject
Environment
Action
Camera
On-screen sound

Write in this order: framing, subject, environment, action, camera, then the sound that happens on screen. Number each shot with [Shot 1], [Shot 2].

overall_soundscape — what the room sounds like

Stops the music burying your dialogue.

Describe ambient sound, foley, and anything a microphone standing in the scene would pick up, in time order. If a spoken line matters, say it sits in front and everything else sits under it.

non_diegetic_music — the score

Music your characters cannot hear. The shortest field.

Give style, instruments, tempo and an emotional curve rather than a track name. A sentence or two is plenty — over-writing it is how people end up fighting their own soundscape.

Assembled prompt, live

integrated_multimodal_description

[Shot 1] Wide shot of a woman in a red coat, wet alley at night, slow push-in, small amplitude.

overall_soundscape

Rain on metal, distant traffic, footsteps in standing water.

non_diegetic_music

Low synth pad, slow pulse, sitting under the dialogue.

0 / 7000

The three fields run in this order and cannot be swapped.

Official output rules

01See it or hear it
02English description, native dialogue
03Keep the order above
04No plot summaries, no stray tags
05Never reorder the fields

Rather not write any of it? The prompt generator fills all three fields from one sentence, and the full guide covers the API too.

Camera and cuts

Camera moves and multi-shot timing in MiniMax H3

Write camera moves and nothing happens. H3 needs three things before it will move the camera at all: type, amplitude and speed.

Type

push-in / pull-out / pan / tilt / orbit / handheld / crane / static

Amplitude

small / medium / strong

Speed

slow / steady / fast

It moves

slow push-in, small amplitude

camera moves

Ignored

slow push-in, small amplitudeThe frame closes on the subject and nothing else moves.
fast aerial push-in, large amplitudeLarge amplitude buys distance covered, not speed of cut.
medium handheld follow, medium amplitudeHandheld is a texture. Say it, or the shot comes back on rails.

Multi-shot timing

Outside the clip — whole generation lost

[Shot 1]

[Shot 2] At 00:04.500

The cut

00:0000:0200:0400:0600:08

8s duration ends

Number your shots and timestamp the cut. Keep the last timestamp inside your chosen duration.

Already have footage you would rather restyle than rewrite? That is video to video.

Dialogue and sound

Dialogue and sound in the same MiniMax H3 prompt

Your characters can talk. The audio comes out of the same pass as the picture, at no extra charge.

The anatomy of a spoken line

Same pass · no extra charge

(S1)speaks softly,[English]Follow the wind, live free.

(S2)replies firmly,[Japanese]分かった。

Two people talking in one shot, each line tagged to a speaker — Two tagged speakers, one take — press for sound

Eleven stable languages

Arabic
Chinese
English
French
German
Italian
Japanese
Korean
Portuguese
Russian
Spanish

The rule people miss

Keep the descriptive parts of your prompt in English, but write each spoken line in its own language.

(S2) replies firmly, [Japanese] 分かった。Right
(S2) replies firmly, [Japanese] Understood.Wrong

Writing a Japanese line in English is the most common reason the delivery comes back wrong.

Need the same face or the same voice across several clips? Use reference to video.

Settings and limits

MiniMax H3 text to video settings, limits, and three things that trip people up

Every value below comes from the official API reference. Three of these rows are where people get caught.

The same shot re-generated at 2K — Same prompt, same seedless run · 2K is re-generated, not upscaled

The same shot generated at 768P — Same prompt, same seedless run · 2K is re-generated, not upscaled

Duration: 4–15 s, whole secondsSeveral sites list 5–15. The API reference accepts 4.
Frame rate: 24 fps
Resolution: 768P or 2K
01Aspect ratio: 21:9 · 16:9 · 4:3 · 1:1 · 3:4 · 9:16 — required
Prompt limit: 7,000 characters
02Frame grid: 17n+5 at 24 fps
03Seed / negative prompt: not exposed
Output: one MP4, H.264 + AAC, audio already muxed

Source: Official · MiniMax API reference

The three things that trip people up

01
Leave the ratio blank and the call fails.
- 21:9
- 16:9
- 4:3
- 1:1
- 3:4
- 9:16
- adaptive
Image to video can stay on adaptive because it has your picture to infer the frame from.
02
Ask for 7 seconds and you may get 6.9.
17n + 5 frames @ 24 fps
Requested
7.0 s
Returned
6.9 s
Duration snaps to the nearest valid block. That is the model, not a bug.
03
No seed. No negative prompt.
7,000 characters here
1,024 tokens on Veo 3.1
Bar length = prompt budget
Control comes from saying what you do want, more precisely. Veo 3.1 gives you 1,024 tokens.

Pricing · 2K on every paid plan · No watermark

What one MiniMax H3 text to video clip costs

A 4-second test costs you less than a 15-second final, because you are billed by output second rather than per attempt. If a generation fails you are not charged at all. Here is what one 8-second clip costs here, next to the two models people usually shortlist against it.

One 8-second clip, list prices

This page · 8s at 768P: $0.64
Veo 3.1 Standard · 1080p: $3.20
Seedance 2.5 · 720p: $1.85

Vendor list prices, checked 2026-08-13. Full pricing.

Same monthly credits either way

FreeNo card
One clip with no account, on the free engine. MiniMax H3 itself is paid.
$0forever
No credit card at any point
Start free
Bot check only — no email
1 clip, no account
Agnes Video V2.0 · 16:9 · 5s · watermarked
Sign in free — 3 clips a day on Agnes, watermarked
Any pack from $9.9 unlocks MiniMax H3 — every mode, 768P and 2K
- MiniMax H3 at native 2KPaid only
- No watermarkPaid only
- One clip, no account
- 1 clip at a time
- Failed clips cost nothing
- Private generation
- Any pack unlocks MiniMax H3 on every mode
Free is a different engine. See plans
Speed & queue
Queue
Free lane
Jobs at once
1
Batch
1
History kept
24h · 7 days signed in
Support
Community
LiteSave 50%
Unlock MiniMax H3 at native 2K. The smallest paid plan — Pro is the one most people pick.
$24.9/mo$49.9
$298.8 billed yearly · Save $300 a year
Choose Lite
7-day refund · cancel anytime
1,290 credits / month · ~22 videos
4s · 768P · text to video
or ~17 at 4s 2K
Same credits monthly or yearly
- MiniMax H3 at native 2K — no watermark
- Dialogue, effects and music in the same file
- Up to 15 seconds15s
- ~22 videos a month at 4s 768P, text to video
- Failed clips cost nothing
- 7-day refund if credits are unused
- 1 job at a time · standard queue
- Batch 2 variants of one prompt
Personal and evaluation use — why
Speed & queue
Queue
Standard
Jobs at once
1
Batch
2
History kept
7 days
Support
Email · 48h
Recommended for most
ProSave 50%
$50 more than Lite. ~125 videos a month, same MiniMax H3, faster queue.
$49.9/mo$99.9
$598.8 billed yearly · Save $600 a year
Choose Pro
7-day refund · cancel anytime
4,990 credits / month · ~125 videos
4s · 768P · text to video
or ~84 at 4s 2K
Same credits monthly or yearly
- Everything in Lite
- ~125 videos a month at 4s 768P, text to video
- Fast queue · 3 jobs at once3×
- Video-to-prompt
- Batch 4 variants of one prompt
- History kept for 7 days
- 7-day refund if credits are unused
Personal and evaluation use — why
Speed & queue
Queue
Fast
Jobs at once
3
Batch
4
History kept
7 days
Support
Email · 12h
StudioSave 50%
A full month of volume, first in line, and a human who answers in 4 hours.
$99.9/mo$199.9
$1,198.8 billed yearly · Save $1,200 a year
Choose Studio
7-day refund · cancel anytime
12,990 credits / month · ~325 videos
4s · 768P · text to video
or ~217 at 4s 2K
Same credits monthly or yearly
- Everything in Pro
- ~325 videos a month at 4s 768P, text to video
- Priority queue · 8 jobs at once8×
- Batch 4 variants of one prompt
- History kept for 7 days
- Priority support in 4 hours
- Lowest credit burn we offer
- 7-day refund if credits are unused
Personal and evaluation use — why
Speed & queue
Queue
Priority — first in line
Jobs at once
8
Batch
4
History kept
7 days
Support
Priority · 4h

Questions people actually ask

MiniMax H3 text to video FAQ

Twelve answers · all visible · nothing collapsed

What is MiniMax H3 text to video?

It turns a written prompt into 4 to 15 seconds of 24 fps video at 768P or 2K, with no source image. Sound arrives in the same MP4.

Do I have to pick an aspect ratio?

Yes. Unlike image to video, adaptive is not available. There is no input image to infer your frame from.

How long can my clip be?

Any whole second from 4 to 15. Several sites list 5 to 15; the API reference accepts 4.

Why did my 7-second request come back at 6.9?

Frame count sits on a 17n+5 grid at 24 fps, so duration snaps to the nearest valid block.

How long can my prompt be?

Up to 7,000 characters. Longer and more specific generally beats short with H3.

Can I use a seed or a negative prompt?

No, neither is exposed. Say what you do want, more precisely, instead.

Can I get 2K?

Yes. 2K re-generates from your original context rather than upscaling pixels, so small text survives.

Does it really generate sound?

Generated in the same pass: dialogue, effects and music. No on/off switch, no separate charge.

Which languages can my characters speak?

Eleven are stable. Write each spoken line in its own language.

Do I need to sign up?

One Agnes Video V2.0 clip with no account. After that, sign in free for 3 clips a day on the free engine, watermarked.

Is my free result watermarked?

Yes. The free Agnes clip is watermarked. MiniMax H3 on a paid plan is not.

Can I use the videos commercially?

Yes, on every plan including the free one. A summary, not legal advice.

One sentence is enough to find out.

No card, no account for one Agnes clip. After that, sign in free for 3 clips a day.

Generate free