SenseTime Releases SenseNova 6.8 Flash Lite Preview with Multimodal Agent Abilities

SenseTime announced today that its SenseNova large model series has introduced a new multimodal agent model, SenseNova 6.8 Flash Lite Preview, now available on the SenseNova Token Plan platform. This model focuses on being lightweight and agile, with capabilities for autonomous planning, continuous execution, multi-agent collaboration, and dynamic error correction. Users only need to set a goal, and the AI completes the entire process independently, delivering end-to-end results.

For long-horizon complex tasks, the model can handle hundreds of steps, spanning phases and lasting hours, while maintaining goals and key facts, and can autonomously backtrack and re-plan upon failure. The main agent can dynamically schedule over a dozen specialized agents to perform tasks such as information retrieval, data analysis, and visual understanding in parallel.

The model natively supports multimodal fusion, processing text, images, charts, documents, videos, and application interfaces. It can create dynamic PPTs, operate browsers and apps, and automate office work. The official versions, 6.8 Flash Lite and the more powerful 6.8 Flash, are expected to be released soon.