MiniMax Hailuo 3.0 is now available on Higgsfield. It's a major step up from Hailuo 2.3, with higher resolution, longer clips, and native sound built into the generation. The model takes images, video, and audio, all feed into the same shot. This guide covers what changed, how to access it, and what it costs.
What Is MiniMax Hailuo 3.0
Hailuo 3.0 is MiniMax's third-generation video model.The biggest change is not only resolution or clip length. H3 can work with text, images, video, and audio together in the same context.
That matters most in what a single generation can actually pull together. A face comes from a reference photo, a camera move comes from a reference video, a voice comes from a reference audio clip, and all three combine into one shot. Sound comes out of that same generation as well, dialogue, sound effects, and ambience built in alongside the picture from the start, not layered on in a separate audio step once the video is already finished.
Hailuo 3.0 vs Hailuo 2.3
Four things changed between versions. Resolution goes from 1080p to 2K, clip length from 10 seconds up to 15-seconds, and audio goes from silent to native stereo generated in the same pass as the video. On Higgsfield, you can add up to 9 images, 3 videos, and 3 audio references to one H3 generation, plus Start/End Frame control on Higgsfield.
Setting | Hailuo 2.3 | Hailuo 3.0 |
|---|---|---|
Resolution | 768p, 1080p | 2K |
Clip length | 768p up to 10 seconds 1080p up to 6 seconds | Up to 15 seconds |
Frame rate | Not specified | 24FPS |
Audio | Silent, no native sound | Native stereo, dialogue, SFX, and ambience |
Input | One image+prompt | Up to 9 images, 3 videos, and 3 audio clips in one generation + prompt |
Aspect ratio | Fixed | 21:9, 16:9, 4:3, 1:1, 3:4, 9:16, or adaptive |
Editing | Not available | Instruction-based editing on a finished clip |
Frame control | Not available | Start/End Frame supported on Higgsfield |
How to Access Hailuo 3.0
MiniMax H3 runs inside Higgsfield, so there is no separate MiniMax account or subscription to manage. Access depends on your Higgsfield plan. It sits alongside Seedance and the rest of the platform's generation models, so switching to H3 for a specific shot is a model selection, not a different account or a new bill.
Full Workflow: Step by Step
Step 1: Open Higgsfield and select Hailuo 3.0 from the model list. It's available in the same generation interface as every other model on the platform.
Step 2: Choose text-to-video or image-to-video, and add your references. Upload up to 9 images, 3 videos, and 3 audio clips, whatever combination the shot needs, a face from a photo, a camera move from a video, a voice from an audio clip.
Step 3: Set Start/End Frame if the shot needs it. This locks the opening and closing frame of the clip directly, available specifically through Higgsfield's implementation of the model.
Step 4: Write the prompt. Describe what happens in the scene, and tell H3 what each reference is for. For example: use Video 1 for the camera movement, Image 2 for the character, and Audio 3 for the voice. Naming the camera move in professional shot language, a slow orbit, a handheld tracking shot, gives a more consistent camera path than describing the scene alone.




