SenseTime (0020.HK) on November 2 launched the official version of SenseNova 6.8 Flash, moving model capabilities from single-turn Q&A toward continuous execution of complex tasks, according to the Shanghai Securities News. The version builds cross-task, cross-session collaboration into the model itself and extends generation scope to static web pages, dynamic content, and design outputs combining aesthetic and practical value. In multiple benchmark evaluations, SenseNova 6.8 Flash surpassed some leading international models in agent collaboration, data analysis and presentation generation scenarios. SenseTime also examined performance in office automation, information retrieval, multi-step collaboration, content generation and long-context understanding. Office scenarios are a key focus: relying on long-term memory and proactive interaction, the model can continuously collaborate around user goals and constraints and identify potentially affected matters when relevant information changes, offering adjustment suggestions. Material cases show the model can link feature changes in a new product version with prior customer needs, proactively flagging that a client who previously declined cooperation over private deployment issues is worth re-engaging. The model extends design generation to Web, SVG, Three.js, Blender, PPT and other formats, supporting dynamic PPTs, dynamic posters and Agentic 3D design, and can plug into workflows such as Figma.