ChainThink report, on April 10, Alibaba has officially confirmed the video generation model HappyHorse-1.0 as an in-house developed product. The model originated from the former Future Life Lab team at Taotian Group, which was reorganized under the newly established Alibaba Token Hub (ATH) business group’s “AI Innovation Division” following Alibaba’s latest structural adjustment.

In anonymous voting conducted on the third-party evaluation platform Artificial Analysis, HappyHorse-1.0 significantly outperformed ByteDance’s Seedance 2.0 and Kuaishou’s Kling 3.0 in pure video generation tasks, while demonstrating comparable performance to Seedance 2.0 in audio-visual unified generation scenarios.

According to insiders close to Alibaba, HappyHorse-1.0 is just one of several multimodal models developed by the team; Alibaba is expected to launch another distinct multimodal model in the near term. Currently, HappyHorse-1.0 remains closed-source, aligning with Alibaba’s recent strategic shift toward proprietary models—since late March, all newly released models by Alibaba have been non-open source.

This accelerated rollout of multimodal models follows ByteDance’s Seedance 2.0 achieving exceptional performance during the 2026 Spring Festival period, which surprised Alibaba internally. Moreover, multimodal generation substantially increases token consumption, directly impacting market share in the MaaS (Model as a Service) sector—according to IDC data, by the first half of 2025, VolcEngine had captured 49.2% of the MaaS market, while Alibaba Cloud held only 27%.