【#Tech24H】SenseTime officially released and open-sourced the new multimodal model architecture NEO, developed in collaboration with the AI Technology Laboratory (S-Lab) at Nanyang Technological University. It is the industry’s first available native multimodal architecture achieving deep-level fusion. The NEO model, based on this architecture, achieved performance comparable to top-tier Vision-Language Models (VLMs) under modular architectures using only 390 million training data points. The model needs just one-tenth the data required by other models with similar performance.