MiniMax H3 — AI Video with Native Audio

MiniMax's open-weight omni-modal video model on Imagenter AI. Text, images, video and audio go in; video with synchronized stereo sound comes out in a single pass. 480p or 768p, 5 to 15 seconds, free trial credits on signup.

Loading...

What Is MiniMax H3?

MiniMax H3 is MiniMax's open-weight, omni-modal video model, released in August 2026. One model reads text, images, video and audio together and returns a clip whose soundtrack is generated with the picture. On Imagenter AI it runs hosted, so there is nothing to download and no GPU to rent.

  • Picture and Sound Together
    Music, dialogue, foley and room tone are generated in the same pass as the video and timed to the action, with no separate audio step.
  • Multimodal References
    Combine up to 9 images, 3 video clips and 3 audio tracks in one generation to carry a subject, a camera move and a voice forward together.
  • Open Weights, Hosted for You
    MiniMax published the weights alongside its API. Running them locally needs a 24GB+ graphics card; here you generate from any browser and pay per clip.

How to Use MiniMax H3

From idea to finished clip in four steps:

MiniMax H3 Features at a Glance

Everything the MiniMax H3 generator supports on Imagenter AI:

Native Stereo Audio

Score, dialogue and ambience generated with the video rather than added after it.

Three Input Modes

Text to Video, Image to Video with first and last frames, and Reference to Video.

9 Images, 3 Videos, 3 Audio Tracks

Reference inputs per generation, cited by number in the prompt.

480p or 768p, 5 to 15 Seconds

Pick the resolution and any whole-second length in that range.

Seven Aspect Ratios

16:9, 9:16, 1:1, 4:3, 3:4, 21:9 ultrawide and 9:21 tall.

Pay Per Second

From 12 credits per second, the estimate is shown before you generate, and failed generations are refunded.

FAQ

FAQs about MiniMax H3

Common questions about running MiniMax H3 on Imagenter AI

1

What is MiniMax H3?

MiniMax H3 is an open-weight, omni-modal AI video model from MiniMax, released in August 2026. It accepts text, image, video and audio inputs and returns video with synchronized stereo audio in a single pass.

2

How much does MiniMax H3 cost?

Text and Image to Video cost 12 credits per second at 480p and 24 credits per second at 768p, so a 5-second 480p clip costs 60 credits. Reference to Video costs 15 or 30 credits per second, plus 6 credits per reference image or audio track and 15 credits per second of reference video. The exact cost is shown before you generate.

3

Is MiniMax H3 free?

Signing up for Imagenter AI is free and new accounts get free trial credits, but they don't cover a MiniMax H3 video: the shortest clip, 5 seconds at 480p, costs 60 credits. Credit packs start at $10 for 1,000 credits, enough for sixteen of those clips. Credits never expire, and failed generations are refunded.

4

MiniMax H3 vs Wan 3.0: which should I use?

Both generate video with native audio from text, images or references. MiniMax H3 costs less per second at its lowest resolution (12 credits at 480p against 15 for Wan 3.0) but tops out at 768p and 15 seconds. Wan 3.0 goes up to 1080p and 30 seconds (30 credits per second at 720p, 60 at 1080p) and accepts more references: 10 images, 5 videos and 5 audio clips against 9, 3 and 3. Pick MiniMax H3 for short, low-cost clips and Wan 3.0 for longer or sharper ones.

5

Does MiniMax H3 generate sound?

Yes. Every clip comes with stereo audio generated alongside the picture. Describe the music, dialogue or ambience you want in the prompt, or supply up to 3 audio tracks in Reference mode.

6

How do references work?

In Reference to Video, upload up to 9 images, 3 video clips (15 seconds combined) and 3 audio tracks, then cite them in the prompt as <Picture 1>, <Video 1> and <Audio 1>. Reference videos are only supported at 480p output.

7

Do I need a GPU to use MiniMax H3?

No. The open weights need a 24GB+ graphics card to run comfortably on your own machine, but Imagenter AI runs MiniMax H3 hosted, so you can generate from any browser.

8

Is Hailuo 2.3 still available?

Hailuo 2.3 has been retired on Imagenter AI. MiniMax H3 is MiniMax's newer model here: it adds native audio, multimodal references and clips up to 15 seconds.