HappyHorse 1.0 Reference

• 🎬 Reference-to-video generation model from Alibaba
• 🖼️ Drives video output from reference image inputs
• 🎯 Part of the HappyHorse 1.0 video generation suite
• 🏢 Built by Alibaba, the Wan/Qwen creator
• 🎨 Tuned for character and subject consistency

Alibaba

Alibaba Group is a Chinese multinational technology company founded in 1999 and headquartered in Hangzhou, Zhejiang. Originally built around e-commerce and cloud computing, Alibaba has become one of the most prolific contributors to open-weight AI research,…

HappyHorse 1.0 Reference is the reference-driven member of Alibaba's HappyHorse 1.0 video generation family, which arrived in April 2026 alongside the text-to-video and image-to-video [[happyhorse-1-0-image-to-video|HappyHorse 1.0]] entry points and the [[happyhorse-1-0-video-to-video|HappyHorse 1.0 Edit]] video-to-video variant. Rather than starting purely from a prompt, this configuration conditions generation on one or more reference inputs, letting creators carry a specific subject, character, or visual style through into the synthesized footage.

Within Alibaba's broader media lineup, HappyHorse sits next to the established Wan video models — including [[wan-2-7-reference-to-video|Wan 2.7 Reference]] — and the Qwen Image family, giving the catalogue multiple routes into AI video. As the current incumbent of the HappyHorse reference-to-video line, it has no newer same-family successor.

It is best suited for workflows where consistency to a provided reference matters most: maintaining a recurring character, matching an existing look, or animating a defined subject across shots rather than generating from scratch.