TMTPOST -- China's generative AI startup HiDream.ai has officially launched HiDream-O1-Image-Pro, a natively unified image foundation model featuring over 200 billion parameters. Built on the proprietary Pixel-level Unified Transformer (UiT) architecture, the model processes raw pixels, text, and conditional tasks within a single shared token space, eliminating traditional external components like Variational Autoencoders (VAEs). Alongside the product debut, HiDream.ai announced the completion of a new nine-figure yuan funding round backed by Co-Stone Capital, Jinpu Investment, Caixin Capital, and Fuju Capital. This marks the company's second major capital injection within a two-week window, following a 500 million yuan round closed in mid-April.