With US restrictions limiting its access to advanced tech, SenseTime is doubling down on open source with a new model optimized to run on Chinese-made chips.SenseTime, a leading Chinese artificial intelligence (AI) firm renowned for its facial recognition technology, has unveiled a groundbreaking open-source model that it asserts can generate and interpret images at a pace far exceeding that of top models developed by its American rivals. The new model, dubbed SenseNova U1, could potentially enable SenseTime to regain its footing in the fiercely competitive Chinese AI landscape, after experiencing a slight dip in its standing among the country’s AI frontrunners.
The unique strength of SenseNova U1 lies in its ability to “read” images without the need to convert them into text first. This innovative approach not only accelerates the process but also significantly reduces the computing power required. “The model’s entire reasoning process is no longer confined to text. It can reason with images as well,” explained Dahua Lin, co-founder and chief scientist at SenseTime, in an interview with WIRED.
Lin, who also holds a professorship in information engineering at the Chinese University of Hong Kong, believes that models capable of processing images directly will pave the way for robots to gain a more profound understanding of the physical world in the future.
In a similar vein to DeepSeek’s latest flagship model, SenseTime asserts that U1 can be powered by Chinese-manufactured chips. “Several Chinese domestic chipmakers have completed compatibility optimization with our new model,” Lin revealed. On the day of the release, ten Chinese chip designers, including Cambricon and Biren Technology, declared their hardware support for U1.
This adaptability is crucial, given that US export controls limit Chinese companies’ access to the world’s most advanced AI chips, particularly those used for training, which are predominantly developed by Western companies like Nvidia. “We will continue to push for training on more diverse chips,” Lin stated. However, he also conceded that SenseTime “may still need to use the best chips to ensure the speed of our iteration.”
SenseTime made U1 available for free on Hugging Face and GitHub, further demonstrating how Chinese companies are emerging as some of the most active contributors to open-source AI.
Established in 2014, SenseTime quickly ascended to global prominence in the field of computer vision, a technology integral to applications such as facial recognition and autonomous driving. However, when AI systems powered by natural language processing, such as ChatGPT, became the tech industry’s latest sensation, SenseTime found it challenging to maintain profitability and fell behind newer Chinese startups like DeepSeek and MiniMax.
By publicly releasing SenseNova-U1 for anyone to use, SenseTime hopes to regain its competitive edge against both domestic and Western AI players. Lin revealed that the company decided last year to concentrate on open-source due to the valuable feedback it receives from researchers, which allows the company to iterate faster. “In this day and age, being open source or closed source is not the winning factor; the speed of iteration is,” Lin clarified.
Embracing open-source also enables SenseTime to continue its collaborations with international researchers, unhindered by geopolitical tensions. The company has faced repeated sanctions from the US government in recent years over allegations that it has been involved in human rights abuses. Despite these challenges, SenseTime remains committed to advancing AI technology and contributing to the global AI community.
