When Alibaba Research released Wan 2.1, it appeared on Hugging Face with a permissive open-source licence and an academic paper. Within two weeks, it had 200,000 downloads and had become the model of choice for professional video AI workflows across Asia's creative industries.
What Wan 2.1 does
Wan 2.1 is a text-to-video and image-to-video generation model. On the metrics that matter for professional use — temporal coherence, motion realism, prompt adherence, and resolution — it outperforms every open-source alternative and competes with the best proprietary models from Runway and Pika at a fraction of the cost.
The Asian creative market response
Adoption has been fastest in South Korea and Japan, where professional animation and video production studios face cost pressure from streaming platforms demanding faster turnaround. Korean webtoon studios have integrated Wan 2.1 into animated adaptation workflows. Japan's advertising industry has integrated it into at least a dozen major production houses.
In China, the model is being used at scale by the gaming industry — particularly for cinematic cutscene production — and by short video platforms Douyin and Kuaishou, which use it for creator-facing AI tools.
The quality bar it has raised
Wan 2.1's release closed the gap between open-source and proprietary video generation. Runway, Pika, and Kling have all released quality improvements since January that appear to be responses to the competitive pressure Wan 2.1 created.
What comes next
Alibaba researchers have published pre-prints suggesting Wan 3.0 will include audio-video synchronisation and improved object permanence across longer sequences. If those capabilities land in an open model, the professional video production market in Asia will see another adoption wave.