SenseTime launches SenseNova 6.7 Flash-Lite, a new lightweight multimodal AI agent model. It uses a native multimodal architecture to directly process complex visuals like charts and layouts, bypassing traditional text conversion. This improves efficiency in office tasks like data analysis and PPT generation, reducing token consumption by about 60% in search scenarios. The company also introduced a free token plan for developers and open-sourced core office skills as SenseNova-Skills on GitHub.