Updated
Updated · WIRED · Apr 29
SenseTime launches open-source SenseNova U1 image model with faster generation and interpretation
Updated
Updated · WIRED · Apr 29

SenseTime launches open-source SenseNova U1 image model with faster generation and interpretation

10 articles · Updated · WIRED · Apr 29
  • SenseNova U1, released for free on Hugging Face and GitHub, is supported by 10 Chinese chipmakers including Cambricon and Biren Technology.
  • The model processes images natively without converting them to text, enabling higher speed and lower computing requirements, and can run on PCs and phones.
  • SenseTime aims to regain its AI leadership and overcome US export controls, with hopes that open-sourcing U1 will accelerate innovation and collaboration despite ongoing US sanctions.
With US chip exports now possible, can SenseTime’s domestic-first AI model truly compete globally?
By open-sourcing its best AI models, is SenseTime building a community or giving away its profits?
As China builds its own AI chips, can new US sanctions on equipment stop its rise?
Will SenseTime's fast-thinking AI finally give robots the human-like vision they need for complex tasks?
SenseTime's new AI generates images at lightning speed, but is fast more valuable than flawless?
SenseTime models now beat GPT-5 on some tasks. Is the West's AI dominance quietly ending?