SenseTime Releases SenseNova 6.7 Flash-Lite Multi-modal Model, Token Plan Free for Limited Time

SenseTime recently launched the new lightweight multi-modal agent model SenseNova 6.7 Flash-Lite, designed for real-world workflows. Using a native multi-modal architecture, it can directly understand complex web layouts, document structures, and financial charts, enabling integrated 'see, think, do' capabilities.

After removing the visual-to-text middle layer, the model has fewer parameters but significantly improved agent capabilities, achieving multiple SOTA results in authoritative benchmark tests for its class. It also reduces Token consumption (60% drop in information search scenarios) and delivers millisecond-level feedback, suitable for high-frequency interactive production environments.

SenseTime simultaneously launched the limited-time free SenseNova Token Plan; developers get 1500 call credits refreshed every 5 hours in the first month. Additionally, the core capabilities of SenseNova-Skills have been open-sourced on GitHub.