Sanctioned Chinese AI Firm SenseTime Releases Image Model Built for Speed
SenseTime, a Chinese The AI company, known for its facial recognition technology, released a new open source model on Tuesday that it says can both generate and interpret images much faster than the best models developed by its U.S. competitors. SenseNova U1 could help the company regain lost ground after losing its place among the major players in China’s AI development race.
The secret of the model lies in its ability to “read” images without translating them into text first, speeding up the process and reducing the computing power required. “The model’s entire reasoning process is no longer limited to text. It can also reason with images,” said Dahua Lin, co-founder and chief scientist of SenseTime, in an interview with WIRED.
Lin, who is also a professor of information engineering at the Chinese University of Hong Kong, says models that can directly process images will allow robots to better understand the physical world in the future.
Like DeepSeek’s latest flagship model, SenseTime claims the U1 can be powered by chips made in China. “Several Chinese chipmakers have finished optimizing compatibility with our new model,” says Lin. On release day, 10 Chinese chip designers, including Cambricon and Biren Technology, announced that their hardware supports U1.
This flexibility is important because U.S. export controls prevent Chinese companies from accessing the world’s most advanced AI chips, especially those used for training, which are currently mainly developed by Western companies like Nvidia. “We will continue to promote training on more different chips,” Lin says. But he also acknowledges that SenseTime “may still need to use the best chips to ensure the speed of our iteration.”
SenseTime released U1 for free on Hugging Face and GitHub, another sign of how Chinese companies are becoming some of the most active contributors to open source AI.
SenseTime was founded in 2014 and has become a global leader in computer vision, used in applications such as facial recognition and autonomous driving. But when ChatGPT and other AI systems powered by natural language processing became the hottest thing in the tech industry, SenseTime started struggling to turn a profit and fell behind new Chinese startups like DeepSeek and MiniMax.
SenseTime hopes that the public release of SenseNova-U1, available to everyone, will help it catch up with domestic and Western AI players. Lin says the company finally made the decision last year to focus on open source because of the helpful feedback it receives from researchers, which allows it to iterate more quickly. “These days, being open source or closed is not the winning factor; speed of iteration is,” says Lin.
Moving to open source also allows SenseTime to continue collaborating with international researchers without the interference of geopolitics. The company has been repeatedly sanctioned by the U.S. government in recent years over allegations that its facial recognition technology helped power surveillance systems used to surveil and detain Uyghurs and other minority groups in China’s Xinjiang region. As a result, American companies cannot invest in SenseTime and sell certain technologies to it without a license. (SenseTime has denied the allegations.)
See clearly
In an accompanying technical report, SenseTime claims that SenseNova-U1 generates higher quality images than all other open source models currently on the market. Its performance is comparable to major Chinese closed-source models such as Alibaba’s Qwen and ByteDance’s Seedream, but it still lags behind industry leaders like GPT-Image-2.0, released just a week ago.
But the main selling point of the model is its ability to generate images much faster than all of these models. It builds on an innovative technical framework called NEO-Unify that SenseTime previewed earlier this year.
