Tag
Russian mathematicians from startup Mostik have developed a method for AI models to communicate via their weights, enhancing efficiency and performance, as demonstrated with GLM-5.2 and Qwen-3.5 models.