Seedance 2.0 Photo: VCG

In a traditional Chinese courtyard, a young Jackie Chan dressed in black engages in a high-speed chase with Jet Li, clad in white. Set to the stirring theme "A Man Should Strengthen Himself", the two leap from courtyard walls into an interior hall, launching into a meticulously choreographed fight packed with close combat and technical precision.

The sequence, however, is not from a newly released action blockbuster. It is a short video generated by Seedance 2.0, the latest AI video model developed by ByteDance, the Chinese parent company of TikTok. The model has attracted widespread attention from global industry professionals. 

Many international media outlets were also quick to take notice following its official release on Thursday.

In an article published the same day, Forbes wrote that the model "nails real world physics and hyper-real video generation" and said "Seedance 2.0 offers a level of creative control that mimics a human director."  

Reuters said in China, the model has been compared to DeepSeek and won praise for its ability to produce cinematic storylines with just a few prompts.

Meanwhile, the Bloomberg highlighted that TikTok owner ByteDance has also won high praise for the performance of its new video model, Seedance 2.0.

While the model has also faced controversy - particularly over its highly realistic rendering and the use of real human likenesses without authorization, an issue common across many video-generation AI systems - the debate surrounding Seedance 2.0 is rapidly expanding beyond technical performance.

Analysts said it may accelerate structural shifts not only in the entertainment industry but also within the video-generation AI sector itself. 

Seedance moment

Within days of trial access being circulated, social media feeds across X were flooded with demonstration clips. Filmmakers, AI engineers and digital artists dissected frame rates, lighting transitions and motion coherence - and many were amazed at how quickly the technology had matured.

Elon Musk, owner of xAI, summed up the accelerating pace of AI development in a brief but telling comment beneath related content: "It's happening fast."

For some in Hollywood, the reaction went beyond curiosity. Filmmaker Charles Curran said on X after testing Seedance 2.0 that with 20 minutes and $60, he successfully generated a trailer for a film featuring characters from the US game Halo. 

Rhett Reese, a Hollywood writer and producer involved in Deadpool 1 & 2 and Deadpool & Wolverine, reposted a Seedance-generated clip depicting a fight between Brad Pitt and Tom Cruise, writing bluntly: "I hate to say it. It's likely over for us." He added: "In next to no time, one person is going to be able to sit at a computer and create a movie indistinguishable from what Hollywood now releases." 

"In the past, we hoped to surpass big companies' models in niche areas through better algorithms," a former film producer now working as a product manager at an AI startup told the Global Times. "Now, Seedance 2.0 suggests ByteDance has achieved a major algorithmic breakthrough. Combined with the vast high-quality training data from Douyin and TikTok, the advantage could be overwhelming for most companies."

Progress nurtured by ecosystem

Just one year after DeepSeek stunned the global tech community in 2025 with a high-performance open-source AI model, China's AI sector has once again demonstrated rapid advancement at the start of 2026. For many observers, the back-to-back breakthroughs are not isolated events, but signals of a gradually maturing innovation ecosystem that is beginning to deliver sustained momentum.

Analysts attribute much of this progress to China's policy orientation, development philosophy in the AI field and strategic resolve - one that combines technological self-reliance with open collaboration.

China has built its AI strategy on technological self-reliance and strength while pursuing open cooperation, Tian Feng, former dean of SenseTime's Intelligence Industry Research Institute, told the Global Times. 

The rapid iteration of domestic large models, he said, is inseparable from the acceleration of open-source ecosystems, which promote technological inclusiveness and deeper industrial applications.

Open-source AI models developed in China are increasing their global influence, breaking overseas technology monopolies while lowering entry barriers for the industry. Behind their growth lie China's institutional strengths, vast market environment, corporate dynamism and long-term strategic focus, Tian said. 

The progress is embodiments of the country's innovation ecosystem under its strategy of achieving greater self-reliance and strength in science and technology, Tian said.

In a previous editorial, the Global Times highlighted China's openness in AI development. "Unlike some countries that engage in technological monopolies in AI development, China's advancements in AI technology not only empower its own economy but also embrace openness and sharing with the world." 

A prior report by CNN also noted this openness, stating that constraints in accessing high-performance chips and capital, as well as China's unique tech ecosystem, have fueled a divergent strategy from the US — making AI models available for public use, or open source. 

Chinese companies have topped global downloads for freely available models and raised substantial funds in market debuts, the report said.