I’ve been using the Composer 1 model for several days and compared it with Claude 4.5. It’s more like a highly capable but rash youngster—when you give it a task to start, it jumps right in without a word, working at lightning speed. However, the results, such as the code it writes, often contain some errors. Overall, Composer 1 is a promising model. If its intelligence can be improved, it could basically rival Claude 4.5.