---
title: "MiniMax H3 Max: Specs, Strengths & Prompt Tips | imaginary.app"
description: "MiniMax H3 Max on fal is fal’s post-trained, fast variant of MiniMax H3, the successor to Hailuo, with sound made with the picture. Specs, tips and FAQs."
url: https://www.imaginary.app/models/minimax-h3
---

Updated November 2026

# MiniMax H3 Max

MiniMax H3 Max, as served by fal, is fal’s post-trained, fast variant of MiniMax H3, the video model MiniMax launched on July 31, 2026 to succeed its Hailuo line. H3 generates sound and picture together, and MiniMax highlights its instruction following and on-screen text.

Key takeaways

- MiniMax H3 launched July 31, 2026 and replaces the Hailuo 02 architecture; H3 Max is one of four H3 API variants.
- MiniMax’s own API lists an H3 Max as “the fast generation variant” at 480P or 768P.
- fal’s H3 Max, which we use, is described by fal as “a post-trained variant of MiniMax H3”, tuned for prompt adherence and aesthetics, with a 1080P option refined from a native 768P source. Neither source says the two are the same model.
- H3 audio is jointly modelled stereo (32 kHz); MiniMax lists 11 dialogue languages.

MiniMax

Generate with MiniMax H3 Max

Write a prompt, pick MiniMax H3 Max, choose your settings and see the credit cost before you start. Failed generations return their credits.

[Open the generator](https://www.imaginary.app/generate)[View Pricing](https://www.imaginary.app/pricing)

Sign in with Google to create an account; buy credits only when you need them.

## On imaginary.app

These are the settings our generator offers for MiniMax H3 Max right now, read from our live model catalog.

- Type

  Video, from a text prompt

- Resolutions

  480p, 768p, 1080p

- Aspect ratios

  16:9, 9:16, 1:1

- Clip lengths

  5 s, 10 s

- Audio

  Always included

- Starting price

  2.97 credits (about $0.63–$0.74) at the lowest settings (480p, 16:9, 5 s, sound included)

The generator shows the exact credits for your settings before you start. USD ranges depend on which [pack or plan](https://www.imaginary.app/pricing) you bought credits with. If a generation fails, or you cancel it before it starts, its credits go back to your balance. Very rarely, when we can't confirm with the provider whether a generation finished, the credits stay reserved while we check, rather than being charged. Models other than Veo 3.1 need an account that has bought a credit pack or plan, and they have daily usage limits, per account and across the site: if one is reached, the generator tells you and you can try again the next day (limits reset at midnight UTC).

## One model for voice, effects and music

MiniMax says H3 models audio jointly with video, with no separation between voice, sound effects and music. Its open-source release puts the audio at 32 kHz stereo, with stable dialogue in 11 languages: Arabic, Chinese, English, French, German, Italian, Japanese, Korean, Portuguese, Russian and Spanish. One prompt can carry a line of dialogue, a door slam and a music bed, already in sync.

## Instruction following and on-screen text

MiniMax’s launch post lists “instruction following, accurate text and brand rendering” among H3’s headline capabilities, and fal says its H3 Max is further tuned “for stronger prompt adherence and better aesthetics”. That suits shots with specific details in a specific order: a sign with a given word, a product on the left, a character who turns before speaking.

## What “Max” means here

MiniMax’s API offers four H3 variants: H3 (768P and 2K), Highspeed, Lite and Max, which it calls “the fast generation variant” at 480P or 768P. fal describes its H3 Max, which we use, as “a post-trained variant of MiniMax H3” optimised for throughput, and adds a 1080P setting made by “latent refinement from a native 768P source”. The two descriptions match on speed, but neither says they are the same model. Either way, H3 Max trades the full H3’s 2K output for speed, and fal’s 1080P is a refinement rather than native generation.

## From Hailuo to H3

H3 succeeds MiniMax’s Hailuo video models. MiniMax says Hailuo 01 built the system from the ground up and Hailuo 02 improved core components, and that H3 set aside the Hailuo 02 architecture because it added unnecessary complexity for handling many tasks in one model.

## MiniMax H3 Max specs from MiniMax

What MiniMax and its API partners document for the model itself. A maker spec is not always a setting our generator offers: what you can choose on imaginary.app is listed in the block above.

| Spec                      | As documented                                                                  | Source                                                                                                                                            |
| ------------------------- | ------------------------------------------------------------------------------ | ------------------------------------------------------------------------------------------------------------------------------------------------- |
| Release                   | MiniMax H3 launched July 31, 2026                                              | [MiniMax API docs — model release notes](https://platform.minimax.io/docs/release-notes/models.md)                                                |
| Lineage                   | Successor to Hailuo 01 and Hailuo 02; H3 drops the Hailuo 02 architecture      | [MiniMax — MiniMax H3 launch blog (Jul 31, 2026)](https://www.minimax.io/blog/minimax-h3)                                                         |
| H3 Max (MiniMax API)      | “The fast generation variant”: 480P or 768P, 5 to 15 seconds on MiniMax’s API  | [MiniMax API docs — video generation V2 (H3, Highspeed, Lite, Max)](https://platform.minimax.io/docs/api-reference/video-generation-v2-create.md) |
| H3 Max (fal endpoint)     | Post-trained by fal; 480P, 768P native, or 1080P refined from 768P; up to 15 s | [fal — MiniMax H3 Max text-to-video (llms.txt schema)](https://fal.ai/models/minimax/h3-max/text-to-video/llms.txt)                               |
| fal release               | August 27, 2026                                                                | [fal — MiniMax H3 Max text-to-video (model page)](https://fal.ai/models/minimax/h3-max/text-to-video)                                             |
| Aspect ratios             | 21:9, 16:9, 4:3, 1:1, 3:4, 9:16                                                | [fal — MiniMax H3 Max text-to-video (llms.txt schema)](https://fal.ai/models/minimax/h3-max/text-to-video/llms.txt)                               |
| Frame rate (open H3 Base) | 24 fps; fal gives no frame rate for H3 Max                                     | [MiniMax — MiniMax H3 open-source release (Aug 3, 2026)](https://www.minimax.io/news/minimax-h3-open-source)                                      |
| Audio (H3)                | Modelled jointly with video: voice, effects and music in one pass              | [MiniMax — MiniMax H3 launch blog (Jul 31, 2026)](https://www.minimax.io/blog/minimax-h3)                                                         |
| Audio (open H3 release)   | 32 kHz stereo; stable dialogue in 11 languages                                 | [MiniMax — MiniMax H3 open-source release (Aug 3, 2026)](https://www.minimax.io/news/minimax-h3-open-source)                                      |
| Open weights              | H3 Base checkpoints (768p) under the MiniMax H3 Community License              | [Hugging Face — MiniMax H3 Community License](https://huggingface.co/MiniMaxAI/MiniMax-H3/blob/main/LICENSE)                                      |

## Who MiniMax H3 Max is for

### Good fit

- Prompts with several specific instructions that all need to land: who, what, where and in which order.
- Short scenes with a sign, label or caption that should read correctly.
- Clips where dialogue, effects and music should arrive together in one pass, in one of 11 languages.
- Iterating quickly at 480P or 768P before picking a take.

### Look elsewhere if

- Native 2K or 4K output; H3 Max tops out at 768P native, with 1080P produced by refinement.
- Silent clips; fal’s H3 Max endpoint has no switch to turn generated audio off.
- Frame-exact edits; fal notes the output can run up to about 0.7 seconds longer than requested.
- Long single takes; MiniMax’s API stops at 15 seconds, and on imaginary.app the available lengths are shown in the “On imaginary.app” panel.

## Prompt tips for MiniMax H3 Max

Spell out the sound as well as the picture: name each voice, effect and music cue you want and when it happens.

Example prompt (written by us)

```
A street musician in a wool coat plays a slow accordion tune under a bridge at dusk as light snow falls. A passing cyclist rings a bell; the musician nods and keeps playing. Soft echo under the bridge, distant traffic.
```

For on-screen text, quote the exact words, say where they sit, and keep it to a word or two.

Example prompt (written by us)

```
Morning in a small bakery: the camera pans across fresh loaves to a chalkboard on the wall that reads "OPEN", then settles on the baker dusting flour off her hands and smiling.
```

Prompts are expanded before generation by default (the “balanced” mode), so lead with the details that must not change, and number the beats if order matters.

Example prompt (written by us)

```
First, a cat leaps onto a windowsill. Second, it knocks a small potted cactus off the edge. Third, it stares at the camera, unbothered. Static wide shot, afternoon light, a soft thud and a meow.
```

Dialogue in any of MiniMax’s 11 listed languages tends to be more stable than other languages, which MiniMax says are “supported to varying degrees”.

## Limitations

- MiniMax says H3’s visual detail “can still be improved in certain scenarios” and that its multimodal understanding has significant room to improve.
- H3 Max’s native ceiling is 768P; 1080P is a refinement of a 768P generation, not a native 1080P or 2K render.
- The open-weight H3 Base licence excludes the European Union, the United Kingdom, South Korea and the United States. Check MiniMax’s and fal’s terms for how they apply to hosted use.
- The clip lengths and resolutions offered on this site are listed in the “On imaginary.app” panel and may be narrower than the maker’s range.
- MiniMax H3 is made by MiniMax; imaginary.app is not affiliated with MiniMax.

## Related models

- [Wan 3.0](https://www.imaginary.app/models/wan-3-0)
- [Seedance 2.5](https://www.imaginary.app/models/seedance-2-5)
- [Kling 3 Pro](https://www.imaginary.app/models/kling-3)
- [Veo 3.1 Fast](https://www.imaginary.app/models/veo-3-1-fast)
- [Every model on imaginary.app](https://www.imaginary.app/models)
- [AI video generator overview](https://www.imaginary.app/ai-video-generator)

## Frequently asked questions

### Is MiniMax H3 Max free?

Not as open weights. MiniMax published the H3 Base checkpoints under the MiniMax H3 Community License, but H3 Max is a hosted variant; on imaginary.app it uses credits.

### MiniMax H3 Max vs MiniMax H3: what is the difference?

On MiniMax’s API, H3 is the flagship variant, generating 768P or 2K, and H3 Max is listed as the fast variant at 480P or 768P. The H3 Max we use is fal’s: fal describes it as a post-trained variant of H3, tuned for prompt adherence and aesthetics, with a 1080P option refined from 768P.

### Is MiniMax H3 the same as Hailuo?

H3 is the successor to MiniMax’s Hailuo video models. MiniMax says it set aside the Hailuo 02 architecture for H3. MiniMax’s own name for it is MiniMax H3, not “Hailuo 3”.

### Does MiniMax H3 Max have audio?

Yes. H3 generates stereo sound jointly with the video, covering voice, effects and music, with stable dialogue in 11 languages. fal’s H3 Max endpoint has no switch to turn audio off.

### What resolution does MiniMax H3 Max output?

MiniMax documents 480P and 768P for H3 Max; fal adds 1080P refined from 768P. Only the full H3 offers 2K. The resolutions offered here are listed in the “On imaginary.app” panel.

### Can I use MiniMax H3 Max videos commercially?

The only MiniMax licence published for H3 covers the open H3 Base weights, and it excludes the European Union, the United Kingdom, South Korea and the United States. H3 Max here is a hosted fal endpoint, so check MiniMax’s terms, fal’s terms and our Terms of Service for how they apply to the clips you make.

## Sources

Prices, model specs and dates on this page were checked on November 10, 2026. Third-party prices change; check the source before you rely on a number.

- [MiniMax — MiniMax H3 launch blog (Jul 31, 2026)](https://www.minimax.io/blog/minimax-h3)
- [MiniMax — MiniMax H3 open-source release (Aug 3, 2026)](https://www.minimax.io/news/minimax-h3-open-source)
- [MiniMax API docs — model release notes](https://platform.minimax.io/docs/release-notes/models.md)
- [MiniMax API docs — video generation V2 (H3, Highspeed, Lite, Max)](https://platform.minimax.io/docs/api-reference/video-generation-v2-create.md)
- [Hugging Face — MiniMax H3 Community License](https://huggingface.co/MiniMaxAI/MiniMax-H3/blob/main/LICENSE)
- [fal — MiniMax H3 Max text-to-video (model page)](https://fal.ai/models/minimax/h3-max/text-to-video)
- [fal — MiniMax H3 Max text-to-video (llms.txt schema)](https://fal.ai/models/minimax/h3-max/text-to-video/llms.txt)
- [imaginary.app pricing](https://www.imaginary.app/pricing)

How we checked: model specs come from MiniMax's and its API partners' pages listed above; the imaginary.app settings and starting price come from our live model catalog. Example prompts are ours, not MiniMax's.

Commercial use: licence and usage terms for MiniMax H3 Max are set by MiniMax. Check them, and our [Terms of Service](https://www.imaginary.app/terms-of-service), before you publish what you make.

MiniMax

## Try MiniMax H3 Max on your own prompt

One account and one credit balance for MiniMax H3 Max, Veo 3.1 and every other video model we offer. The price shows before you generate.

[Open the generator](https://www.imaginary.app/generate)[View Pricing](https://www.imaginary.app/pricing)

Sign in with Google to create an account; buy credits only when you need them.
