SenseTime releases open-source image model that runs on Chinese chips

SenseTime, a Chinese AI company known for its facial recognition technology, has released SenseNova U1, an open-source model that it claims can generate and interpret images far faster than top models from US competitors. The model's key innovation is its ability to process images directly without translating them to text first, which speeds up the process and reduces computing power requirements. SenseTime says the model's entire reasoning process is no longer limited to text and can reason with images as well. Like DeepSeek's latest flagship model, SenseNova U1 can run on Chinese-made chips. On release day, 10 Chinese chip designers including Cambricon and Biren Technology announced their hardware supports the model. This flexibility matters because US export controls restrict Chinese firms from accessing advanced AI chips primarily developed by Western companies like Nvidia. SenseTime released U1 for free on Hugging Face and GitHub. The company says it hopes the public release will help it catch up with both domestic and Western AI players after falling behind newer Chinese startups like DeepSeek and MiniMax. SenseTime was founded in 2014 and became a world leader in computer vision. The company has been sanctioned repeatedly by the US government over allegations that its facial recognition technology helped power surveillance systems in China's Xinjiang region, which SenseTime has denied.