MiniMaxH3 Video Model Tested: Native 2K Resolution & 12 References

azed_ai · x · 2026-07-31

A user tested the MiniMaxH3 model from Hailuo AI and praised the consistency of its generation results.

Key specs of the model include: 15-second video generation; native resolution of 2560 × 1440 (2K); and as an omni model, it supports up to 12 cross-modal references, encompassing images, videos, and audio.

Related event: MiniMax H3 Video Model Impresses in Comprehensive Tests(57 posts)→

Original post →

More from Multimodal

Multimodal channel →