Developed by the Shanghai AI Laboratory with academic and startup partners, Lumina2 was released in January 2025, setting the technological precedent that Z-Image, released in November 2025, later refined. Lumina 2 is a 2.6-billion-parameter AI model for generating images from text, built on a unified "single-stream" architecture that processes text and images as one continuous data flow for perfect synchronization. Z-Image by Alibaba, shares this core design, making them close rivals that balance compact size and speed for home computers. While Lumina2 handles basic text-in-image tasks adequately, Z-Image excels specifically at complex, accurate typography—especially bilingual English and Chinese text—making it superior for professional graphic renderings.

Recently I used the Lumina 2 ALL-in-One model, packaged and redistibuted by Comfy Org Hugging Face Repo, that has both Text Encoder & VAE baked into just a single file. After seeing how good this is I felt it should be known, used and praised by more folks so I mirrored it here on CivitAI. I wanna see some good LORAs trained for this. This is a reupload of resource from this Hugging Face Repo link -

https://huggingface.co/Comfy-Org/Lumina_Image_2.0_Repackaged/tree/main/all_in_one