Skip to content
AI IntelligenceAug 10, 2026AI Intelligence
Article

MiniMax released H3 as open weights, with text, image, video, and audio fused as context and native stereo audio output up to 15 seconds at 2K.

The 768p base model can run on consumer GPUs, making multimodal workflows more accessible.

Data Cube AI EditorialSource: Pandaily
01

Source Brief

MiniMax released H3 as open weights, with text, image, video, and audio fused as context and native stereo audio output up to 15 seconds at 2K. The 768p base model can run on consumer GPUs, making multimodal workflows more accessible.