“In our pursuit to push the boundaries of SenseNova’s capabilities, SenseTime remains guided by the Scaling Law as we build upon our large [language] model based on this three-tier architecture: knowledge, reasoning and execution,” said Li, referring to expanding the LLM’s size by increasing its parameter count to boost its learning and generalisation capacity.

Given the same knowledge input, SenseNova 5.0 provides better comprehension, summarisation, and question and answers to support vertical applications such as in the education and the content industries, according to a SenseTime statement. The updated LLM is also touted to have “best-in-class mathematical, coding and reasoning capabilities, providing a solid foundation for applications in finance and data analysis”.

SenseNova 5.0 also supports high-definition image parsing and understanding, as well as text-to-image generation, according to SenseTime. In addition, the LLM’s latest iteration also extracts complex data across documents and summarises answers to questions, “offering strong multimodal interaction capability”.

The updated LLM, according to SenseTime, had undergone more than 10 terabytes of token training, covering a large amount of synthetic data, which has boosted its context window coverage – referring to the swathe of text that an AI model can process during conversation with users – to about 200,000.