Zao Dian AI

Zao Dian AI, launched by Alibaba’s Quark team after the 2025 Apsara Conference, focuses on AI-generated images and videos. The image module leverages the Midjourney V7 model with localized optimizations, offering styles like 3D, children's illustration, spring portrait, along with size, smart retouching, and keyword settings. The video module runs on the Tongyi One 2.5 model, capable of producing up to 10-second audio-visual clips across realistic and anime styles. Its strengths lie in a user-friendly interface and a rich inspiration feed organized by categories (AI effects, short dramas, wallpapers, etc.), which lowers the entry barrier for novices. Backed by Alibaba’s Tongyi model family, it shows decent Mandarin comprehension and stylized output. However, current evidence indicates that the Midjourney V7 backend sometimes misinterprets Chinese prompts and may inadvertently reproduce web-crawled watermarks, reducing reliability. Public information is limited to early reviews and media summaries; there is no official data on pricing, subscription plans, or long-term availability, so the total cost of ownership and service continuity are unclear.