HiDream-O1-Image: A Natively Unified Image Generative Foundation Model with Pixel-level Unified Transformer

HiDream-O1-Image presents end-to-end pixel-space Diffusion Transformer unifying image generation, text encoding, and task conditioning into single token space, eliminating external VAE and text encoder dependencies.