OpenClips.AI
Pricing
OpenClips.AI
MiniMax · video model

MiniMax H3: brief it like a director

Combines text, images, video and audio in a single context.

Generate Video

Enable JavaScript for the full composer, or continue in the app.

What are you making?

Switching filters the roster and rebuilds the parameter row.

Model

The model reconfigures everything else. Invalid controls simply vanish.

Aspect

Where is this clip going to live? Real shapes, true proportions.

Duration

Pick a length — the price updates as you do.

Attach product, avatar or reference

Attachments are content, not settings — they go in the strip, never into your prompt.

Recent

Your library

Share this clip

Every link carries your ?ref=NOVA21 code — shares earn you credits.

What's new in MiniMax H3

01

Every modality, one context

Text, reference images, video clips and audio — combined in a single brief.

02

2K output

Generate at 2K resolution for delivery that survives the big screen.

03

Style transfer from footage

Show it reference footage; H3 pulls motion and tone, not just content.

MiniMax H3 specs

StatusLive in the composer
InputText + image + video + audio, single context
Max resolution2K
Max duration10 seconds
FamilyHailuo (MiniMax)

Made with MiniMax H3

“A paper boat sails down a city gutter at golden hour, macro lens”

MINIMAX H3

“Extreme macro: espresso crema swirling, steam catching golden light”

MINIMAX H3

MiniMax H3, one paragraph deep

Most models read a prompt; MiniMax H3 reads a production brief — text, reference images, reference footage and audio notes fused in a single context, rendered to 2K. Mood boards become motion. “This product, moving like this footage, feeling like this track” is a one-generation brief, not a week of iteration.

Who H3 is for

Agencies translating client mood boards into motion. Creators remixing their own footage with AI extensions. Anyone whose brief lives in more than one modality. it sits in flagship territory — pair it with Hailuo 02 for cheap drafts of the same idea, and run the final against Seedance 2.5 when image-reference precision is the decider. One composer, one wallet, every take side by side — that’s the whole point.

Frequently asked questions

MiniMax H3 is the newest Hailuo-generation flagship: a multimodal video model that reads text, images, video clips and audio in one context to generate 2K video. It runs on OpenClips per render.

H3 for multimodal briefs (mixed references incl. footage and audio notes); Seedance 2.5 for maximum image-reference control and 30s takes. Brief both — takes land side by side.

Hailuo is the family, MiniMax the vendor; H3 is the current flagship version. On OpenClips the family page is /models/hailuo, with H3 and Hailuo 02 as its live versions.

Priced per render — after sign-in, the Generate button shows the exact credits before you commit. One wallet across all 17 model families.