Alibaba’s Happy Horse-1.0 Secures Top Position in LM Arena’s Blind Test Rankings for Video Editing
10 hour ago / Read about 0 minute
Author:小编   

LM Arena, a globally recognized platform for evaluating AI models, recently unveiled its latest rankings in the Video Edit category. Alibaba’s ATH business group’s Happy Horse-1.0 model clinched the top spot with an impressive Elo score of 1299, outperforming leading international models such as Grok. This achievement marks the first time a Chinese AI team has claimed the number one position in this specific evaluation category, indicating that China’s video generation and editing technology has reached the pinnacle of global excellence.

LM Arena ensures fairness and accuracy in its evaluations by employing an anonymous comparative voting system alongside the Elo rating mechanism. The Happy Horse-1.0 model stands out for its exceptional capabilities, including high-fidelity portrait rendering, seamless motion control, and precise video editing. Prior to this, the HappyHorse series had already led the rankings in both text-to-video and image-to-video categories during anonymous testing on Artificial Analysis’s Video Arena leaderboard.

Experts highlight that Happy Horse-1.0’s victory not only underscores Alibaba’s foundational innovation in the realm of multimodal large models but also signifies a broader trend: AI video technology is evolving towards “precise editing and controllable production.” As a result, the global AI video industry’s competitive landscape is swiftly shifting towards specialized vertical operational scenarios.