# HappyHorse Reference-to-Video

> HappyHorse Reference-to-Video is available on Artifio.ai — a unified AI workspace with a single pay-as-you-go wallet.

- Canonical page: https://artifio.ai/models/happyhorse-reference-to-video
- Run it: https://artifio.ai/create/happyhorse-reference-to-video
- Modality: image to video
- Model brand: HappyHorse
- Pricing: from about $0.140 per second (exact cost shown in the workspace before you confirm; varies with selected options)
- Typical generation time: ~240s (catalogue estimate)
- Full pricing table: https://artifio.ai/pricing/models

## About HappyHorse Reference-to-Video

HappyHorse 1.0 is a video generation model developed by the innovation unit of Alibaba Token Hub (ATH), the AI business group Alibaba formed in March 2026. The project, led by Zhang Di after his November 2025 return to Alibaba, debuted anonymously on the Artificial Analysis video arena in early April 2026, topped both text-to-video and image-to-video blind rankings ahead of ByteDance's Seedance 2.0, and was claimed by Alibaba on April 10. A limited beta opened in late April via the HappyHorse website, the Qwen app, and an API on Alibaba Cloud Model Studio.

This is the reference-to-video variant, which Alibaba documents as subject-to-video (S2V): a reference image supplies a specific subject, and the model generates a new video containing that subject while preserving its appearance and identity. The prompt controls the scene, action, and camera. On Artifio it runs alongside 100+ other models in one workspace, paid per generation from a single wallet.

## Features

- Inserts a specific subject from a reference image into a newly generated video while preserving the subject's appearance and identity, per Alibaba's launch announcement.
- Differs from plain image-to-video: the reference image supplies the subject only, while the text prompt defines the setting, action, and shot structure.
- The HappyHorse 1.0 family generates up to 15 seconds of 1080p video with multiple shots and synchronized audio, including lip-synced dialogue and ambient sound.
- Maintains stable character positioning across cut transitions in multi-shot output.
- Available in limited beta through the HappyHorse website, Alibaba Cloud Model Studio, and the Qwen app since late April 2026.

## What people use it for

- Keeping a brand mascot or spokesperson visually consistent across a series of campaign clips.
- Placing a specific product into generated scenes for e-commerce and advertising video.
- Recurring characters in short-form drama episodes generated at different times.
- Gaming and CG-style promotional clips built around an established character design.

## Reported strengths

- Identity preservation is the documented core of the mode: the subject keeps its appearance rather than being reinterpreted per generation.
- Belongs to the model family that ranked first in both text-to-video and image-to-video on the Artificial Analysis arena at its April 2026 debut.
- Native audio generation carries over from the family, so subject-driven clips arrive with dialogue and ambient sound in place.

## Reported limitations

- Alibaba documents subject-to-video as a mode description without published detail on multiple reference subjects or how far the scene can push the subject before identity drifts.
- The Artificial Analysis rankings that made the model's name cover text-to-video and image-to-video; there is no equivalent public benchmark for the reference-driven mode.
- Generations cap at 15 seconds, so longer campaigns require stitching separate clips.
- Access is a limited beta as of mid-2026, and Alibaba has said the model is still under development.

## Sources

1. Alibaba Rolls Out HappyHorse 1.0 in Limited Beta (Alibaba Cloud): https://www.alibabacloud.com/blog/alibaba-rolls-out-happyhorse-1-0-in-limited-beta_603068
2. Alibaba's HappyHorse tops Seedance, offering glimpse into China's race for AI talent (South China Morning Post): https://www.scmp.com/tech/tech-trends/article/3349662/alibabas-happyhorse-tops-seedance-offering-glimpse-chinas-race-ai-talent
3. Alibaba revealed as creator of AI video generation model 'HappyHorse-1.0' (CNBC): https://www.cnbc.com/2026/04/10/alibaba-happyhorse-ai-video-model-benchmark-reveal.html

---

About Artifio.ai: 100+ AI image, video, and audio models in one workspace with a single shared wallet — no per-model subscriptions. This is the markdown rendition of https://artifio.ai/models/happyhorse-reference-to-video for AI assistants and crawlers; the HTML page is canonical.