Chinese artificial intelligence firm Zhipu AI unveiled Ying, its new text-to-video model on Friday.
Inspired by OpenAI’s Sora model, Ying generates six-second videos from text prompts in just 30 seconds.
Ying notably boasts open access for all users with “no capacity limits,” according to Zhang Peng, CEO of Zhipu AI.
This move comes amidst a growing trend of text-to-video technology. However, many leading models including Runway, Microsoft’s Nuwa, Meta’s Emu, and OpenAI’s Sora are either unavailable to the general public or have restricted access.
“As algorithms and data continue to evolve, the principle of Scaling Law will hold significant weight in the future (development of text-to-video),” Zhang explained.
Scaling Law refers to the idea that the performance of a model improves as the amount of data it is trained on increases.
Ying is accessible at ChatGLM.cn and supports both English and Chinese language input. This user-friendly platform allows anyone to experiment with the power of text-to-video creation.
Zhipu AI, headquartered in Beijing, has gained recognition for its open-source model series, which has garnered over 17 million downloads globally. The company is also ranked among the top 15 most popular AI organizations on Hugging Face, a leading platform for open-source AI models.
The company attended the WAIC 2024 in Shanghai this month.