Alibaba has officially taken credit for the creation of 'HappyHorse'.

This AI video model, once a mysterious 'dark horse', recently surged to the top of the rankings on the authoritative AI evaluation platform, Artificial Analysis, outperforming the much-hyped ByteDance Seedance 2.0 and Kuaishou Kling 3.0.

As of now, HappyHorse-1.0 leads the pack in text-to-video and image-to-video (audio-free) models, particularly excelling in pure video generation sans audio. When it comes to audio-visual integrated generation models with sound, it stands toe-to-toe with Seedance 2.0.

Suddenly, the AI community was abuzz with speculation: Which tech behemoth was behind this 'horse'? Alibaba, ByteDance, or Kuaishou?

HappyHorse: An Alibaba Creation

In fact, its naming follows a discernible pattern. A straightforward interpretation, based solely on the words themselves, uncovers two key terms: Happy and Horse. Since the official Chinese name has yet to be announced, it's often directly translated as 'Happy Horse' or 'Joyful Horse'. When Jack Ma was at the helm, he proposed the 'double H' strategy, symbolizing Health and Happiness.

The mystery surrounding its ownership was finally unraveled after nearly a week. On April 10, the official social media account 'HappyHorse_AI' was launched, with its inaugural post explicitly stating that HappyHorse is an internal testing product of Alibaba's ATH Innovation Business Unit. It also dispelled rumors about a fake official website and revealed that the product is still undergoing refinement and has yet to be officially launched.

Image: Alibaba 'claims' HappyHorse. Screenshot by Tang Chen

Subsequently, Alibaba's official Weibo account reposted the announcement, confirming its Alibaba lineage—a 'horse' that belongs to the Ma family. Further updates indicate that HappyHorse-1.0 is slated to open its API on April 30. Meanwhile, Alibaba is also set to launch another multimodal model, distinct from HappyHorse-1.0.

The debut of HappyHorse follows a familiar playbook. Xiaomi's MiMo-V2-Pro and Zhipu's GLM-5 previously employed similar 'blind box' tactics for their rankings. This strategy—anonymous participation followed by official claims—tests the model's inherent strength and the confidence of the AI company behind it.

Leveraging suspense and mystery, beyond the brand's inherent allure, while achieving tangible results in blind testing, offers far greater cost-effectiveness than traditional product launches. For major AI players, as long as the model's capabilities are robust, even if some flaws are exposed, market and user expectations can still be effectively managed.

Beyond marketing considerations, HappyHorse's technological prowess withstands scrutiny. Based on currently available technical analyses and user feedback from public channels, it adopts a unified modeling approach at the architectural level, utilizing a pure self-attention single-stream Transformer with approximately 15 billion parameters. Text, video, and audio tokens are placed in the same sequence for joint modeling, enabling native synchronous generation of audio and video.

This approach diverges from the prevalent 'video generation + audio post-processing' splice scheme used by mainstream video models, sidestepping the 'uncanny valley' effect of mismatched audio and video in current AI video generation. This means the model 'comprehends' the timing, instructions, and quality requirements of the video while generating the visuals, achieving precise alignment of human figures, sound, and scene.

The direct benefit is 'high efficiency and low consumption'. HappyHorse generates a 5-second 1080p video on a single H100 GPU in about 38 seconds and supports lip-syncing in seven languages. For AI video creators or users, this translates to halving costs while producing higher-quality video content.

Of course, as an exploratory product from ATH, HappyHorse is not a jack-of-all-trades. Current evaluations suggest it resembles a 'photographer' skilled at capturing aesthetically pleasing empty shots rather than a 'director' capable of handling complex narratives. When dealing with intricate physical movements or long-term logical sequences, the model may still exhibit issues like distorted actions or reduced coherence.

Nevertheless, this does not detract from its status as one of the world's strongest open-source video models today. Especially in scenarios like e-commerce, immersive short videos, and AI-generated comics, its practical value for Alibaba and even the broader AI industry extends beyond mere technology.

The 'Token Economy' Monetization Window Has Opened

In my opinion, HappyHorse signifies a 'capability breakthrough' for Alibaba's AI. Prior to this, Alibaba's AI endeavors have been gaining traction on both the B2B and B2C fronts. In terms of models, Qwen 3.6 Plus recently topped OpenRouter's global weekly large model call volume rankings.

The driving force behind this momentum is Alibaba's comprehensive AI strategy over the past two years, including organizational reforms and the construction of 'Tongyunge' (a term referring to Alibaba's full-stack AI capabilities), which have now entered a rapid 'monetization' phase.

In 2023, Eddie Wu assumed the role of Alibaba Group's CEO, partnering with Joe Tsai to make 'AI-driven' one of Alibaba's top priorities. Over the next two-plus years, he systematically overhauled Alibaba's AI efforts.

At the 2025 Apsara Conference, Wu outlined a long-term strategy for Alibaba centered on artificial superintelligence (ASI). He explicitly stated that AGI is merely the starting point for AI development; the ultimate goal is to achieve self-iterating superintelligence that surpasses human capabilities in every aspect.

To this end, Alibaba has made the boldest AI infrastructure investment among domestic AI companies, planning to invest at least RMB 380 billion over the next three years. This sum exceeds Alibaba's total investment in cloud and AI infrastructure over the past decade.

Entering 2026, Alibaba's AI progress has noticeably picked up speed. On March 16 this year, Alibaba CEO Eddie Wu personally spearheaded the establishment of the ATH (Alibaba Token Hub) Business Group, integrating scattered resources such as the Tongyi Lab, MaaS business line, and Qianwen Business Unit. The objective is clear: create tokens, distribute tokens, and apply tokens.

Shortly after, on April 8, Alibaba further established a Group Technology Committee, with Eddie Wu at the helm and members including Jingren Zhou, Zeming Wu, and Feifei Li. The Tongyi Lab was upgraded to the Tongyi Large Model Business Unit, headed by Jingren Zhou.

Within just one month, Alibaba consolidated its AI technical prowess and resources through two organizational adjustments, fully committing them to the AI battlefield.

Especially with the establishment of the Technology Committee, Eddie Wu dissected Alibaba's AI capabilities, previously scattered across different business units, into three distinct technical paths: Jingren Zhou oversees models, Feifei Li handles cloud and AI infrastructure, and Zeming Wu manages the group's business technology platform and AI inference platform.

This model-to-application integration also mirrors Eddie Wu's assessment of the current AI landscape. During Alibaba's Q3 FY2026 earnings call, he stated, 'From the second half of 2025 to early 2026, we have witnessed AI entering a new era driven by agentic capabilities. The key difference from early AI stages lies in the seamless integration between models and applications.'

'Guangzi Xingqiu' (Photon Planet) remarked that Alibaba's strategy can be viewed as a 'triangular framework'. A robust infrastructure forms the bedrock, a unified organizational structure serves as the hub, and a thriving application ecosystem acts as the outlet. This interlocking system of 'chips-cloud-models-applications' represents the systemic advantage Alibaba aims to establish in the AI era—a core competitiveness difficult for rivals to replicate.

Meanwhile, Alibaba Cloud announced price hikes starting April 18, with computing power cards increasing by 5% to 34% and storage by 30%. This reflects both the supply-demand imbalance on the demand side and Alibaba Cloud's pricing leverage on the supply side.

Organizational reforms, model 'breakthroughs', and business acceleration—Alibaba is responding to new trends in the AI industry: the monetization window for the 'Token economy' has swung open, and only collective efforts can seize the initiative.

The backdrop is that tokens are emerging as the new hard currency, ushering in a new era where 'token supremacy' reigns. In March 2026, the average daily token usage of China's AI large models surpassed 140 trillion.

Sora serves as a cautionary tale under these new 'rules of the game'. OpenAI reluctantly shut it down, demonstrating that the era of competing solely on model parameters and technical sophistication has passed. The 'brute force' approach of simply scaling up is no longer tenable. The next competitive focus will shift from 'whether models work' to 'how well they work' and 'how cost-effective they are'. The key lies in achieving a sustainable commercialization loop.

According to Alibaba's plans, its goal is to surpass USD 100 billion in annual revenue from cloud and AI commercialization within the next five years. This necessitates Alibaba's AI to demonstrate not just technical prowess in 'having models' but also commercial viability in generating revenue and value.

Eddie Wu's pressure stems from this: over the next five years, Alibaba must achieve a compound annual growth rate exceeding 40% in cloud and AI commercialization revenue. Alibaba needs to bolster its multimodal capabilities to sell more tokens. Compared to pure text-based chat, multimodal generation consumes significantly more tokens, profoundly impacting cloud computing services' market share in MaaS (Model as a Service).

Looking back, HappyHorse represents Alibaba's calculated strategy for the 'Token economy'. It brings Alibaba's AI strengths and pressures to the fore, driving deep integration between technology and real-world scenarios, unlocking synergies across Alibaba's business ecosystem, and serving growth and monetization objectives.

After all, in the AI race, many horses gallop swiftly, but only those carrying provisions home—achieving self-sufficiency—can endure longer and go farther in the AI marathon.

So, will the pressure now shift to ByteDance, Kuaishou, or Tencent?

References:

Guangzi Xingqiu (Photon Planet), 'Alibaba's Organizational Upgrade: Decisive Battle in the AI Arena'

Time Weekly, 'Alibaba's AI Acceleration'