On April 27, Alibaba's video generation model HappyHorse 1.0 officially started gray testing, first landing in the Qwen App. Users can experience the beta version through the HappyHorse button at the bottom of the homepage. This move signals Alibaba's video generation layout transitioning from technical reserves to product validation stage.
Technical highlight: native architecture for joint audio-video generation
Unlike the serial solution in most markets of first generating video then dubbing, HappyHorse 1.0 adopts a native multimodal architecture, achieving joint audio-video generation. This means visual frames and audio are collaboratively optimized within the model, theoretically bringing better audio-visual sync effects and more consistent style expression. From the officially disclosed information, the model supports 15-second multi-shot narrative, multi-aspect-ratio adaptation, and 1080P super-resolution output, with picture quality, narrative ability, character performance, and style diversity as key optimization directions.
Clear positioning: commercialization path for content production
Unlike many tech-demo-style video models, HappyHorse has clear commercial-scenario positioning from the start: advertising, e-commerce, short drama, social media creative. The list price published on Qwen's official website also reveals clear commercialization intent — 720P video at 0.9 yuan/second, 1080P at 1.6 yuan/second. This pricing is below most similar products on the market, showing Alibaba's strategy of entering the B-end market with cost-effectiveness.
Why this is worth attention
Video generation models are evolving from "can generate" to "can tell stories." HappyHorse's multi-shot narrative capability is a key step — not just generating a segment of footage, but generating short video units with camera language and narrative logic. For short drama and ad production, this means the usability of AI-assisted content production is significantly improved.
Of course, gray-test-stage capability ≠ production-stage experience. The 15-second time cap, character consistency in multi-shot scenes, actual generation speed, are all core metrics to watch subsequently. Whether Alibaba can leverage Qwen's traffic entry and e-commerce ecosystem to iterate quickly will determine whether HappyHorse is a flash in the pan or genuinely shakes up the industry landscape.