On April 10, a rising star emerged in the domestic multimodal large model arena. The anonymous model "HappyHorse-1.0" topped the authoritative evaluation for text-to-video (without audio) with an Elo score of 1332, surpassing the second-place ByteDance's Dreamina Seedance2.0 by nearly 60 points, drawing significant industry attention. According to the latest market updates, HappyHorse has launched accounts on overseas social platforms, and its first follower is Alibaba Group, further confirming the strong R&D capabilities behind this model.

It is reported that HappyHorse is suspected to originate from the "Future Life Lab" under Alibaba's TaoTian Group, led by Zhang Di's team. Currently, the lab has completed its organizational restructuring and has been officially transferred to the AI Innovation Division under the ATH business group. The unexpected rise of HappyHorse not only demonstrates Alibaba's deep technical foundation in multimodal generation but also reflects the trend of internal incubation mechanisms in major tech companies toward more independent and efficient organizational structures.

Against the backdrop of intense global competition in AI video generation, HappyHorse's leading position is not just about parameters or rankings, but also marks that domestic models have already demonstrated the capability to compete at the world's top level in core aspects such as video narrative logic and visual quality. As the model moves from anonymity to commercialization, Alibaba's AI strategy under the ATH business group will become increasingly clear, potentially reshaping the current competitive landscape of video large models.