Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
SenseTime, from China An AI company known for its services face recognition technology, released a new open source model on Tuesday that claims to be able to create and translate images much faster than the top models produced by US competitors. SenseNova U1 can help the company recover the lost space after leaving its premises leading players in China’s AI development competition.
The secret sauce of the model is its ability to “read” images without translating them to write first, speeding up the process and reducing the required computer power. “All model thinking is no longer based on text. It can also think with images,” Dahua Lin, cofounder and chief scientist at SenseTime, said in an interview with WIRED.
Lin, who is also a professor of information technology at the Chinese University of Hong Kong, says models that can directly generate images will help robots better understand the world in the future.
Like the latest version of DeepSeek, SenseTime says the U1 can be powered by chips made in China. “Several Chinese chip makers have completed integration with our new model,” says Lin. On the day of the release, 10 Chinese chip manufacturers, including Cambricon and Biren Technology, announced that their devices support the U1.
That flexibility is important because US export control they prevent Chinese companies from getting the world’s most advanced AI chips, especially those used in education, which are currently made by Western companies such as Nvidia. “We will continue to push education on different chips,” says Lin. But he also admits that SenseTime “may still need to use better chips to ensure the speed of our replication.”
SenseTime has released U1 for free on Hugging Face and GitHub, another sign of how Chinese companies are becoming some of the biggest 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 driving. But as ChatGPT and other natural language-driven AI systems became the hottest thing in the tech industry, SenseTime struggled to make a profit and fell behind new Chinese startups like DeepSeek and MiniMax.
SenseTime says it hopes that releasing the SenseNova-U1 to the public for everyone to use will help it become more compatible 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 gets from researchers, which helps the company iterate quickly. “Today, being open source or closed source is not a success factor; the speed of iteration is,” explains Lin.
Going open also allows SenseTime to continue to engage with international researchers without the distraction of geopolitics. The company has been repeatedly criticized by the US government in recent years for claims that its facial recognition technology has aided surveillance efforts used to monitor and detain Uyghurs and other minority groups in China’s Xinjiang region. As a result, US companies are prohibited from investing in SenseTime and selling other technologies without permission. (SenseTime has denied these claims.)
Seeing Clearly
In the following technical report, SenseTime claims that the SenseNova-U1 produces better images than any other open source model on the market. Its performance is comparable to leading Chinese blockchain models like Alibaba’s Qwen and ByteDance’s Seedream, but it still lags behind industry leaders like GPT-Image-2.0, which came out last week.
But the main selling point of this model is that it can produce images faster than all other models. It relies on a technology called NEO-Unify that SenseTime previewed earlier this year.
