Alibaba has officially &#34;claimed&#34; the mysterious video generation model HappyHorse, which has sparked widespread speculation in the AI circle. This move marks Alibaba's strongest achievement in the video AI track to date and serves as the latest footnote to its comprehensive acceleration of its AI strategy.
Alibaba ATH confirmed on April 10 that HappyHorse is a model developed by its innovative business division, currently in the internal testing phase, and will open API access in the near future; it is reported that the API opening date is set for April 30.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/01914c23-b5ef-4d54-9fd9-148d414a8fd0.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/01914c23-b5ef-4d54-9fd9-148d414a8fd0.png"/>
The model previously topped the text-to-video leaderboard of Artificial Analysis, a world-renowned AI evaluation platform, in anonymous form, pushing ByteDance's Seedance 2.0 to second place, triggering large-scale speculation in the AI circle and investor community.
The emergence of HappyHorse directly boosted Alibaba's stock price—during the period the news fermented, Alibaba's stock price rose by more than 4% intraday. Jefferies analyst Thomas Chong wrote in a research report that HappyHorse is &#34;a success&#34; for Alibaba.
<h2>Topping the Leaderboard, Crushing Peers</h2>
The debut of HappyHorse-1.0 was quite rare: no press conference, no technical blog, and no corporate endorsement, directly appearing anonymously on the Artificial Analysis Video Arena leaderboard and taking the top spot.
Leaderboard data shows that in the text-to-video (excluding audio) track, HappyHorse-1.0's Elo score reached 1379, 106 points higher than Seedance 2.0, and also higher than mainstream products such as Kunlun Wanwei's SkyReels V4 and Kuaishou's Kling AI 3.0. In the image-to-video (excluding audio) track, the model scored 1411 points, surpassing Seedance 2.0's 1356 points and xAI's grok-imagine-video's 1331 points. According to reports from Wallstreet News, the score in the image-to-video track set a new record in the leaderboard's history.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/0f16d61b-37a0-41d3-b333-60817bd7d4d3.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/0f16d61b-37a0-41d3-b333-60817bd7d4d3.png"/>
Notably, in the &#34;video + audio&#34; comprehensive ranking including native sound effects, HappyHorse ranked second, slightly below Seedance 2.0.
Artificial Analysis's leaderboard uses a blind test mechanism, letting real global users compare and choose between generated videos without knowing the model's source, and aggregating results through an Elo rating system similar to chess. Compared to traditional evaluation methods where manufacturers self-report results, this mechanism is considered to provide a more objective signal of model quality. However, since the sample size for new models is relatively limited, their Elo score volatility is usually greater than that of older models; furthermore, Elo is a relative rating, and the industry generally believes a gap of about 60 points represents a stable win-rate advantage.
<h2>Controversy and Skepticism Coexist</h2>
The sudden rise of HappyHorse is not without controversy. Some observers pointed out that in Artificial Analysis's blind test samples, portrait generation and talking-head content account for more than 60%, and HappyHorse has a natural advantage in portrait scenarios, which may to some extent cause a gap between the evaluation score and actual comprehensive capabilities.
On social platform X, discussants are divided into two camps: skeptics believe there is still a visible gap between HappyHorse and Seedance 2.0 in character detail and dynamic coherence; supporters place high hopes on its potential, especially looking forward to its ability to solve the industry pain point of image quality consistency in multi-shot sequences.
<h2>ATH Business Group: The New Engine for Alibaba's AI Acceleration</h2>
The Alibaba ATH Business Group behind HappyHorse, full name Alibaba Token Hub Business Group, was officially established on March 16 this year. According to The Paper, Alibaba CEO Eddie Wu issued an internal announcement declaring the establishment of the business group with the core goal of &#34;creating tokens, delivering tokens, and applying tokens,&#34; with Eddie Wu directly in charge.
On April 8, Eddie Wu issued another internal letter announcing the establishment of a Technology Committee at the group level, with himself serving as the chairman and members including Zhou Jingren, Wu Zeming, and Fei-Fei Li. Among them, Zhou Jingren serves as the Chief AI Architect, Fei-Fei Li is responsible for Alibaba Cloud technology and AI cloud infrastructure construction, and Wu Zeming is responsible for the group's business technology platform and AI inference platform construction. Meanwhile, the Tongyi Laboratory was upgraded to the Tongyi Large Model Business Unit, headed by Zhou Jingren.
According to Bloomberg, Alibaba's previous video generator under the Wan brand had a highest ranking of around 20th on the Artificial Analysis leaderboard. The topping of HappyHorse represents a significant leap in Alibaba's video AI capabilities. Alibaba stated that 2026 is a key year for its accelerated AI development. From the establishment of the ATH Business Group to the recent Qwen 3.6 Plus model claiming the top spot in OpenRouter's weekly global large model call volume, Alibaba's recent moves in the AI field have been continuous.
Currently, video generation is one of the most fiercely contested tracks for AI developers and one of the few directions proven to achieve reliable commercialization. With OpenAI exiting the track last month, opportunities for Chinese companies have further expanded, and most of the main products on the Artificial Analysis leaderboard are currently from Chinese companies.
Alibaba ATH stated that HappyHorse is part of its plan to explore entirely new interaction methods in the AI era, and more products will be launched in succession. The opening of the API will allow external developers to access the model, further testing its commercial potential in real-world application scenarios.

BABA

09988

BABX

KBAB

BABO

Alibaba's ATH Business Group has officially confirmed that it developed the mysterious video generation model HappyHorse. The model topped Artificial Analysis's global text-to-video leaderboard in anonymous form with an Elo score of 1379, surpassing ByteDance's Seedance 2.0, while its image-to-video score set a new historical record for the leaderboard. The news boosted Alibaba's stock price by more than 4% intraday. HappyHorse will open its API on April 30

- Alibaba has officially claimed the AI video generation model HappyHorse, marking a significant achievement in video AI technology and reflecting its accelerated AI strategy.  
- HappyHorse was developed by Alibaba’s ATH division, currently in internal testing, with an API launch scheduled for April 30.  
- Despite some controversy over its evaluation metrics, HappyHorse's performance on the Artificial Analysis platform has led to a rise in Alibaba’s stock price and underscores its enhanced capabilities in the competitive AI landscape.

Wallstreetcn

Alibaba

BABA-W

BABA 2x Long Daily ETF - GraniteShares

KraneShares 2x Long BABA Daily ETF

YieldMax BABA Option Income Strategy ETF

Crushing Seedance 2.0, Alibaba Announces: "HappyHorse" Video Model Undergoing Internal Testing