On November 6, 2025, the Meituan LongCat team proudly announced the launch of UNO-Bench, a cutting-edge, all-encompassing multimodal one-stop evaluation benchmark. This innovative benchmark is designed to precisely assess a model's proficiency in both unimodal and multimodal understanding within a unified framework. Notably, it marks the first instance of validating the 'Composition Law' for multimodal large models, thereby presenting the industry with a fresh and insightful analytical approach.
