0
Applied AI·July 31, 2026·1 min read

MiniMax releases H3, a video model that generates up to 15-second clips in 2K resolution with native stereo sound, and plans to release its weights within days

Share

A 2K, 15‑second video model with native stereo sound and near‑term weight release pushes high‑fidelity multimodal generation into the open ecosystem. This compresses the moat around proprietary video models and gives media, gaming, and marketing teams a reason to revisit their build‑vs‑buy assumptions on synthetic content pipelines.