MiniMax has open-weighted H3-Image, a 33.1 billion parameter diffusion transformer for text-to-image and editing, produc...

MiniMax has open-weighted H3-Image, a 33.1 billion parameter diffusion transformer for text-to-image and editing, producing roughly 16 megapixel output at 4096 by 4096 in under 10 seconds. It is distilled to six sampling steps without classifier-free guidance and accepts up to 16 reference images. The community licence excludes the EU, UK, US and South Korea. https://pandaily.com/minimax-h3-image-open-weights-16mp-text-to-image-editing-six-step #China #Tech #AI