(Yicai) March 18 -- China's Tencent Holdings revealed its latest progress in artificial intelligence by unveiling Follow-Your-Click, an image-to-video model developed together with two top universities, but OpenAI's Sora is still unbeatable, according to industry insiders.

Tencent, Tsinghua University, and the Hong Kong University of Science and Technology recently released the source code of the tool on Microsoft’s open-source code website GitHub.

The demo video shows how users can click on a certain area of an image and make objects move for a few seconds with prompts such as "turn head," "smile," "sad," and "launch." For example, if pressing "sad" on an image of a person, the character will bow their head and cover their face to cry.

However, Chinese companies have not yet been able to create a match for Sora, OpenAI's text-to-video generator, industry insiders told Yicai.

Moreover, there is a noticeable gap between domestic tools and Midjourney, an American image-generation tool similar to OpenAI's DALL-E, a technical professional who studies AI applications said. “Domestic image generation tools are roughly one version behind their overseas counterparts, with a gap of about six months to a year.”

However, startups have no shortage of cash or ambition as several local developers of large language models have recently secured financing and claimed that they are catching up with Sora.

For example, Tang Jiayu, chief executive of Shengshu-AI, said that the startup expects to catch up with the current version of Sora by December. Shengshu-AI recently completed a new round of funding totaling hundreds of millions of yuan.

Shengshu-AI's peer AISphere recently wrapped up its Series A1 round of financing worth 100 million yuan. Founder Wang Changhu said that the text-to-video startup plans to concentrate its resources to surpass the latest version of Sora in three to six months.

Even ByteDance is joining the hype as the parent of TikTok launched MagicVideo-V2, its text-to-video model, in January. The Beijing-based firm is developing a tool called Boximator to allow users to reshape their videos by moving objects in isolation.

Editor: Emmi Laine