HappyHorse 1.0 Reference

• 🎬 Reference-guided video generation from Alibaba
• 🖼️ Turns reference images into coherent motion clips
• 🎯 Part of the HappyHorse 1.0 video family
• 🎨 Built for stylized, character-consistent video output
• 🔄 Now joined by a newer 1.1 release

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-to-video member of Alibaba's HappyHorse 1.0 generation, released in April 2026. Where the line's text-to-video and image-to-video variants start from prompts or single stills, this Reference model conditions generation on supplied reference imagery — useful for keeping a character, style, or subject consistent across the synthesized clip. It launched alongside the HappyHorse 1.0 text-to-video and image-to-video entries, plus the HappyHorse 1.0 Edit video-to-video tool, forming a complete first-generation video toolkit.

Within Alibaba's broader catalogue, HappyHorse sits next to the larger Wan video family and the Qwen text, image, and embedding lines, giving the provider coverage across most generative modalities. This particular model is the prior generation of the Reference line: it has since been joined by HappyHorse 1.1 Reference (released June 2026), which carries the same reference-driven approach forward.

It's best suited to creators who need video that faithfully reflects a provided visual reference — maintaining a recurring character or aesthetic — rather than improvising purely from a text prompt. Users may still choose 1.0 deliberately for its specific output behavior, while those wanting the latest refinements can step up to the 1.1 release.